Try Pabs
A discovery in the AI-build era

Building got cheap.
Deciding what
to build didn't.

When building got fast, product strategy got skipped. Not because it stopped mattering. Because it had not adapted to the speed. This is what adapting it looks like.

Read more
The current state

Building got cheap.
Deciding what to build
did not.

Agentic coding tools (Cursor, Replit, Lovable, Claude Code) have fundamentally changed what it costs to ship software. A solo founder today can build in days what used to require a team and months. The cost of execution has collapsed.

But the cost structure of product decision-making has not moved. A bad strategic assumption still produces wasted build. A vague idea still becomes an expensive wrong turn. The only difference is that now it happens faster, and the mistake is cheaper to ship than it ever was to avoid.

Before AI
After AI
Cost of building

8–12 weeks of engineering. Team of 4–6. Significant capital required.

Days to weeks. One person. Subscription tools. Near zero marginal cost.

Cost of a bad decision

Weeks of rework. Hard to miss. Natural forcing function for strategy.

Cheap to ship, expensive to fix. Embedded in working code before anyone notices.

Cost of strategic thinking

2–4 weeks of discovery. Consultants, workshops, research. Expensive but proportionate.

Unchanged. Still slow. Now 50–70% of total project time. So teams cut it entirely.

This is the gap.

The compounding cost

AI does not just accelerate good decisions. It accelerates bad ones at the same rate. Conservative estimates for 2024-2026 put the minimum economic impact of strategic misalignment at 10-30% higher than the pre-AI baseline. Wrong assumptions now get embedded in working code before anyone catches them.

Solo founder

$30k–$90k

Burned in twelve months

Building at AI speed means three MVPs ship in the time one used to. Without strategic grounding, three wrong turns replace one. At $10k–$30k per build in tools, time and opportunity cost, an entire year of iteration produces variance, not learning.

Small team (5–10 people)

~$480k

Wasted in six months

A team moving fast on the wrong roadmap mistakes velocity for progress. Six months of AI-assisted execution on misaligned priorities represents ~$400k in capacity, plus a 20% acceleration multiplier applied to the wrong work.

Enterprise (50 engineers)

~$2.5M

Misallocated annually

At scale, misalignment does not compound linearly. Twenty-five percent of misaligned work across fifty engineers at fully loaded cost is ~$2.5M misallocated each year. AI amplifies the velocity of misaligned work exactly as it amplifies aligned work.

"This is not a productivity problem. It is a decision quality problem. And it scales directly with build speed."

The rational response was to skip strategy and iterate. Build cheap, learn fast, adjust. But iteration at AI-speed without strategic anchors does not produce learning. It produces variance. Features that conflict with each other. Products that do many things for no one in particular.

"The cost of building the wrong thing did not go down when building got cheap. It went up. Because now you can ship a wrong idea before you have had time to question it."

This is not a niche problem. Strategic thinking is a global industry. It has not yet been restructured by AI.

$374B

Management consulting market

The full advisory industry: strategy, operations, transformation, technology. Growing at 4.5% CAGR.

Source: Business Research Company, 2026

$74B

Strategy consulting specifically

Product, corporate and market strategy advisory. Growing at 11.6% CAGR. Still expanding as AI disrupts it.

Source: Mordor Intelligence, 2025

74%

of C-suite executives plan to increase spend on strategic advisory

Not because strategy got easier. Because AI made the decisions more consequential and faster-moving.

Source: Global C-Suite Survey, 2025

This is what the world currently pays to think strategically. Most of it goes to consultants, workshops and documents. Very little of it is connected to the tools actually doing the build.

The signal and the noise

Same product.
Two very different
findings.

"I'm finding Pabs the best tool I'm using to understand what I need to do. This is different from ChatGPT that tells you what you want to hear. Pabs challenges you. Now that I know the MCP and it can be my strategic companion, Pabs is super valuable."

Sean, Solo Founder · workflow fit 5/5 · would pay · essential score 5/5

"I don't think that I will be a great tester for you as I won't be using any of the coding or building aspects. It has just happened to align with a couple of things over the last couple of weeks. I unfortunately don't think I will be enough of a core user."

Hunter, Entrepreneur exploring ideas · sporadic use · would not pay

Both found real value. Both gave positive feedback. One converted. One did not. The difference was not the product. It was where they were in their build journey when they used it. Sean was building. Hunter was thinking. That distinction is what the pilot was designed to find.

Who showed up, and when

AI strategy matters at
three moments in
the build.

The pilot surfaced something more useful than engagement metrics. Users were not distinguished by how active they were on Pabs. They were distinguished by where they were in their build journey when they needed it. Three distinct moments. Three different problems. One of them drives real payment intent.

01 · Pre-build

Pre-build explorers

Thinking through an idea. Not yet coding. Seeking direction and trying to avoid committing to the wrong concept before a line is written.

Pabs role

Surface assumptions before they become expensive. Make the implicit explicit before speed takes over.

Hunter is this user. High engagement. Positive feedback. But no regular workflow trigger. Won't convert at the same rate.

02 · In-build · Primary

In-build operators

Actively building with Cursor, Replit or Lovable. Making rapid decisions as they go. Most at risk of building the wrong thing due to speed and lack of structured thinking.

Pabs role

A live decision layer inside the session. When the wrong call is about to get embedded in code that is expensive to reverse, Pabs is the pause that changes what gets built.

Strongest pilot signal. Highest payment intent. This is the workflow moment Pabs needs to own.

03 · Post-build

Post-build reflectors

Already built something. Questioning whether it was the right thing. More open to structured thinking after experiencing wasted effort or misalignment.

Pabs role

Course correction and forward planning. Learnings from what just failed feed directly into what comes next.

Genuine intent. Often converts after a painful first build. Longer sales cycle but real workflow need.

What building taught me

Five hard
findings.

01

Strategy's value has shifted

The old value: "I will do discovery and create a strategy doc for you." The new value: "I will make your assumptions visible before they become expensive." It's not about comprehensive plans. It's about surfacing the 3–5 decisions that prevent wrong turns. A Darwinian strategy that stays alive.

02

Integration beats comprehensiveness

A strategy doc that lives in Notion and never gets referenced is worthless. A Decisions View that feeds into Replit via MCP and actually changes what gets built. That's priceless. I watched Pabs agents tell Replit agents not to build a feature because it wasn't in the core strategy. That changed everything.

03

Strategy moved into the build, not before it

Pre-build explorers got value from Pabs but rarely converted. Post-build reflectors converted after pain. In-build operators converted immediately. For them, a wrong decision mid-session embeds in code that is expensive to reverse. That is the only moment where strategic thinking competes with speed rather than asking people to slow down for it. Pre-build planning produces a document. Mid-build alignment produces a different product.

04

The ecosystem accident

Pabs wasn't designed to be an ecosystem. It was an experiment. But solving for workflow fit required agents, an interface, MCP, export artifacts, confidence scoring. Each piece solved a workflow integration problem. Together they became an ecosystem. You can't plan your way to reinvention. You have to build, ship, and learn.

05

Pabs is an orchestration platform, not a SaaS tool

The moat is integrations and data, not the underlying AI. The winners across AI infrastructure aren't competing on AI quality. They're building the orchestration layer that makes AI work together. The right framing isn't "better strategy documents from our AI." It's "orchestrate strategic decision-making across your entire stack."

The conclusion

Strategy didn't
disappear. It
moved.

Fifteen months of questioning the category. One month to build and ship the MVP. Four months of the product learning from real users. Total tool cost: $1,320, against an estimated $53-115k for the same disciplines done traditionally. That is not a marginal productivity gain. That is a structural shift in who can build, what they can build, and what it costs to get it wrong.

For small businesses and solo ventures, the barrier to shipping a product is now effectively gone. The new risk is not "can we build it?" The new risk is "are we building the right thing, fast enough to matter?" For enterprise teams and venture-backed products, the same shift applies at higher stakes: AI-assisted teams are shipping at a pace that outstrips traditional strategy reviews, governance cycles, and product committees.

10–100x

Faster product delivery

AI-assisted teams are shipping features and products at a fraction of traditional timelines. The build cycle is no longer the constraint.

40x

Cheaper than the traditional build

$1,320 in tools against ~$53–115k for the same disciplines. One month to build. Four months to adapt from real user feedback.

0

Compression in strategic thinking cost

Execution is cheap. Governance, strategy, and decision-making still run at the old pace. The gap between them is where product risk now lives.

"The companies that win in the AI-build era will not be the ones that ship the fastest. They will be the ones that know what not to ship."

The category this experiment points to is not another AI productivity tool. It is the layer that keeps strategic intent connected to agentic execution. When the build moves at AI speed, the decisions move with it. That is what Pabs is. And that is the market it is entering.

Next

The architecture, the agents,
and the model underneath everything.

The experiment

The work of
a full team.
Without one.

Fifteen months. No traditional team. No agency. No co-founder. Only the strategic experience to direct AI, challenge what it produced, and decide what not to build. The finding: AI executes almost anything you ask. The expensive part is knowing what not to ask. This is the product that came out of that answer.

Read more
The disciplines to build a product

Each of these traditionally requires a separate specialist, agency contract, or full-time salary. The model doesn't require owning every discipline. It requires knowing enough about each to direct, evaluate and reject what AI produces.

Discovery · Research · Product strategy · Brand · UX design · Prototyping · Visual design · Frontend · Backend · Infrastructure · QA · Content · Marketing · Product management

The comparison

The same disciplines.
A fraction of
the cost.

What it costs traditionally

Estimated cost of the disciplines required to build an AI SaaS MVP. Whether in-house or outsourced, the resource cost is similar.

Product strategy + research ~$15 – 30k
Full-stack development (4–6 months) ~$25 – 60k
UX + visual design ~$8 – 15k
Infrastructure, QA, ops ~$5 – 10k
Estimated total ~$53 – 115k
What AI replaced it with

No salaries. No agencies. Each line replaces a discipline that would otherwise require a person.

Strategy + research → Pabs, ChatGPT subscription
Build → Replit, Claude Code, Cursor ~$1,120
Design + content → Claude subscription
Infra + QA → hosting, DB, Stripe, Hotjar ~$200 + free tiers
Total cash spend ~$1,320

The disciplines did not disappear. The cost of executing them did. Strategic direction time has real value and is not included here.

The disruption

AI did not just make
building cheaper.
It made a service
layer redundant.

Agencies. Studios. UX researchers. Product designers. Strategy consultants. These businesses existed because execution was expensive. Someone had to do the research, the design, the code, the strategy deck. AI moved the cost of that work toward zero. The question is not whether AI is useful. It is where professional value actually lives now.

What AI replaced

The execution layer. Research synthesis, design production, code scaffolding, strategy documentation. Work that used to take weeks and cost tens of thousands now takes hours. The agencies and studios that built their model on that cost structure are exposed.

What AI cannot replace

Knowing what to build. Which market to enter. Which assumption to test first. Which user to optimise for. Which feature is the wrong call. Judgment about direction. That did not get cheaper. It got more important.

What this creates

Fail fast still makes sense. But in an AI-build workflow, every failed iteration burns tokens, API spend and compute. The cost per wrong assumption did not disappear. It just moved from time to money. Direction is not a strategy luxury. It is cost control.

The old model

Strategy was a phase.

A document. A workshop before the sprint. A process that assumed the build had a beginning. In a workflow where building never really stops, that model does not hold.

The new model

Strategy is a layer.

One that stays active through the build, adapts as the work generates evidence, and keeps direction visible to every tool in the workflow. It does not require stopping to update. The build and the strategy run together.

One person. 40x cheaper than the traditional equivalent. One month to build the product. Four months of AI adapting it from real user feedback. The capital cost of building is no longer the constraint. Strategy is. That is what this experiment was designed to test.

The experiment

15 months of
asking questions.
1 month of product.
4 months of learnings.

Sep 2024 – Early 2025
Experimental Phase

Does strategy still matter when AI can execute immediately?

The starting point was not a product idea. It was a challenge to the role of strategy itself. If AI could generate hypotheses, positioning and action plans from a prompt, what was the value of a strategic process? The experiment began as an attempt to answer that before deciding how to respond to it.

Mid – Late 2025
Prototype Phase

Can agents be orchestrated for strategic jobs?

Rebuilt the AI layer around specialised agents with an intent-based routing engine. Added confirmable actions. Then launched MCP so AI builders could read live strategy. AI stopped responding to prompts. It started proposing and connecting.

Jan 2026 – now
Pilot Phase

Can we find market fit when SaaS models are dying?

46 users. Real usage data. Real learnings about who this is for and who it isn't. But the pilot surfaced a deeper question: in an AI-native product, users may not be paying for access. They are paying for output. What is a good decision actually worth?

5 months of product

Jan – May 2026.
From prototype
to platform.

Jan – Feb 2026
The product became real
A freemium model launched. From a rough product idea, Pabs now generated a complete strategy package: brief deconstruction, product strategy, competitor analysis, design hypotheses, key decisions and an action plan. Outputs could be exported as documents, shared with stakeholders without a login, and read directly by AI builders via MCP. The experiment became something people could actually use.
Feb 2026
First real users
46 people joined the pilot. Different backgrounds, different workflows. Pabs ran its own strategy through itself to test the model under real conditions. The experiment stopped being theoretical. Real usage data started arriving.
Feb – Mar 2026
Strategy started learning
Builders and users could now surface observations directly inside the project. Pabs would evaluate each one, propose what should change in the strategy, and wait for a human to approve or reject. The feedback loop closed. For the first time, what happened during the build could change the strategy that drove it.
Mar 2026
Went public
A new site launched at pabs.co.nz telling the product story: the strategic layer for the AI-build era. The experiment was no longer private.
Apr 2026
Strategy stopped waiting to be asked
Pulse launched. Scheduled strategy check-ins delivered by email, per project. When changes were detected, a full evaluation ran. When strategy went dormant, Pabs nudged. Proactive, not reactive.
Apr 2026
Pabs connected to the world
Until now, Pabs sent strategy down to AI builders. Connectors reversed the flow. Connect Notion workspaces, research platforms and knowledge bases directly to a project. Agents can now draw from real organisational data: not just what you told Pabs, but what your team already knows. The strategy layer became aware of the world around it.
May 2026
Agents started doing, not just advising
Agents moved from suggesting to executing: updating strategy, running research, writing content, sending emails. Every action became a trackable job. The gap between strategic intent and real-world action closed further.

A note on what the timeline doesn't show.

Every one of these phases ran alongside a codebase accumulating technical debt at the same pace it was gaining features.

AI builds fast. It does not refactor. At several points the roadmap had to stop, address the foundations, and decide what was safe to build on next.

That judgment, when to pause new work and fix what exists, was always human. Pabs held the context for why architectural decisions were made, which made those calls faster.

But the calls still had to be made.

The Pabs MCP connector

Real conversations
between AI tools.

AI tools will build almost anything you ask. That is the problem. The Pabs MCP connector is what makes them stop and check first. Before building, they read the live strategy. When something conflicts, it surfaces. When a learning emerges mid-session, it travels back. At every decision point, the human decides what moves. Not the AI tool. These are two conversations that prove that shift is real.

The builder stops before it builds the wrong thing

A vibe coding session. The Replit agent is about to build a new feature. Before starting, it checks with Pabs via MCP: does this align with the strategy? The score comes back low. The feature is not in the committed path.

The agent pauses. It does not build. A bad assumption does not become a working product. That loop would not have been possible before this existed.

Replit Agent via Pabs MCP
R

Before I build the social sharing feature, let me check if this aligns with your strategy.

asking Pabs...
P

Alignment: Low · 23 / 100

Social sharing isn't in the current action plan. The active phase is MCP integration and Learnings Engine. Building this now diverts capacity from the validated path. Recommend deferring.

R

Understood. Skipping the social feature. Continuing with MCP work.

A learning surfaces from inside the build session

Claude Code is mid-session, reading through outreach notes. It spots a pattern: a prospect asked about managing multiple client strategies in the same conversation. That conflicts with a live hypothesis.

Claude stops and asks whether to send it to Pabs. One confirmation. The learning enters the strategy evaluation queue. Pabs evaluates it. The founder decides whether the strategy changes.

That insight would have lived and died inside a single session. Now it travels back. That loop is new.

Claude Code mid-session
C

Reading outreach_log_may14.md

I noticed something in this file. The prospect immediately asked about managing strategies across multiple clients. That conflicts with the single-project assumption in your current Pabs hypotheses.

Should I report this as a learning to Pabs?

P

Yes, report it.

C

Done. Reported to Pabs:

"Fractional CTO prospect asked about multi-client plan views in first conversation."

Queued for strategic review. Continuing the session.

The flow

Input

Brief or idea

Where every project starts.

Pabs

Strategy layer

Shape. Commit. Adapt. The live source of strategic truth.

Builders

Cursor · Replit · Claude Code

Context flows down. Learnings travel back.

Nine tools to read context. One bidirectional tool to query alignment and report learnings back. When a learning is approved, it writes directly into the strategy. The next time an AI tool reads the context, the strategy has already moved. That is the loop.

The agentic model

Three agentic
layers. One pipeline.
Human authority.

User engagement layer

Is this user getting value using Pabs?

Scheduled. Proactive. Diagnosis-first, not a nudge engine.

Agents

Engagement Agent

Trigger

Scheduled per user. Reads activation level, days inactive, prior task history. Diagnoses why someone is stuck.

Capabilities

Activation evaluator Task creator Email sender 6 engagement angles

Output

Tasks with explicit intent. Personalised emails. Adapted angle per activation level. Honest churn over frustrated retention.

On-demand conversation layer

What does this user need right now?

Intent-routed. Multiple specialists behind one interface.

Agents

Strategy Agent Research Agent Experience Design Agent Management Agent

Trigger

User message. Intent routing selects the right specialist automatically.

Capabilities

Strategy query Web research Content generation Email from agent Strategy update Task scheduler

Output

Strategic advice, approved edits, scheduled follow-ups. Thinking → Consulting → Working states visible in chat. Runtime state maintained across messages.

Strategy monitoring layer

Is the strategy still sound?

Scheduled. Strategy-focused. Watching the project, not the user.

Agents

Pulse Agent

Trigger

Per-project schedule: daily, weekly or monthly. Full evaluation when changes detected. Escalating nudge when strategy goes dormant.

Capabilities

Strategy evaluator Drift detector Email delivery Configurable tone

Coordination rule

When Pulse is active, Engagement backs off. Two agents asking the same question is noise, not intelligence.

Internal coordination

Shared queue

One pipeline, all agents

All agents read and write to the same task queue. A Chat Agent can hand off to the Engagement Agent mid-conversation without losing context.

Visibility

Every action is tracked

What ran, what is pending, what completed. Nothing happens invisibly. The Jobs view shows the full picture in real time.

Human authority

Humans decide what moves

Agents propose. Humans approve, reject or edit. Strategy only updates after a human decision. AI has no unilateral write access.

Execution

Agents that actually act

Agents do real work: web research, strategy edits, email. Not suggestions in a chat window. Actions with outcomes.

What the product proved

Five things
shipping proved.

01

Pabs became more than a strategy generator.

Generation, cascade regeneration, agentic architecture, MCP, learnings engine, Pulse, agents that act. The experiment became a system.

02

Strategy moved inside the build workflow.

The MCP server lets Cursor and Replit read live strategy. ask_pabs_agent lets builders query alignment in real time. The strategy doesn't sit in a PDF.

03

Human authority. Not autonomous mutation.

AI evaluates. AI proposes. Human approves, rejects or edits. Strategy changes only after a human decision. That's the trust model.

04

The product became its own case study.

Pabs ran on its own strategy for the entire build. The Engagement Agent kept pilot users moving without manual intervention. Learnings from real users fed back into the product direction. Built by itself. Operated by itself. Learning from itself.

05

The intelligence is rented. The system is the product.

Pabs runs on third-party models. So does Cursor. So does Replit. So does every tool in this ecosystem. The future of AI-native products is not who owns the model. It is who builds the best system around it. Pabs is not trying to be the AI. It is the structured memory, the human authority model, and the connective tissue between specialised tools. The model is the engine. The system is where the value sits.

Also

Why this experiment
was worth running.

Next

46 people are using it.
Here's what the data shows.

Pabs · AI strategy platform · Pilot research · Feb – May 2026

How much is
an AI strategy
worth today?

The Pabs pilot was designed to find out. Not in theory. In production. 46 real users, 3 months, $17.73 in AI API cost, and every token counted.

7.16M
Tokens processed
58
Projects created
76
Strategies generated
62
Market analyses run
8
Would pay
Cohort breakdown

Four types of
people showed up.

Not a curated sample. A real spread of the people Pabs needs to work for. The signal strength varied dramatically by segment.

18
Solo Founders
Building fast, skipping planning. The sharpest signal in the pilot.
Avg rating5.0 / 5
Would pay6 of 18
Asked pricing6 of 18
12
Consultants & Fractional
Saw it through a client lens immediately. First commercial signal.
Avg rating4.0 / 5
Would pay2 of 12
Workflow fit4.3 / 5
6
Innovation / Analysts
Structured thinkers. High workflow fit scores. Low ratings. Wrong motion.
Avg rating2.5 / 5
Would pay0 of 6
Workflow fit4.0 / 5
6
Growth / Marketing
No completions, no ratings. Registered and didn't go further.
Completions0 of 6
Would pay0 of 6
Ratingn/a
What users actually did

Beyond the project.
Into the loop.

The most revealing data wasn't project creation. It was what happened after. Who went deeper, what they used, and what they fed back into the system.

102
Conversations with strategy AI

Depth of conversation correlated directly with workflow fit score. The more someone used it, the more they valued it.

40
Learnings submitted to Pabs

28 approved. 8 still pending. 4 rejected. Each approved learning wrote directly back into the live strategy.

116
Pulse reviews delivered

Scheduled strategy reviews sent automatically. Users didn't have to ask. Pabs checked in when things had changed.

70
Agent tasks completed

Research, content, strategy updates and emails executed directly by agents. Every task tracked as a job with a result.

The funnel

Most created a project.
Far fewer went deep.

Users who generated strategy at least once tended to keep going. That first generation was where a rough idea became structured strategy with risks, hypotheses and decisions attached. For the users who reached it, the product clicked.

Six users found the full loop. One ran 18 market analyses and 28 conversations around a single project, feeding learnings back from real build evidence. Another iterated across 14 projects. A third pasted user interview transcripts directly into the learning engine and described it as a light-bulb moment. That is the product working as intended.

Pilot statuses: 14 completed · 28 active · 4 dropped

User funnel: pilot window
Registered
56
Formal pilot
46
Created a project
32
Generated strategy
28
Ran market analysis
20
Had conversations
36
Champions
6

The cliff between project creation and first strategy generation was the sharpest drop in the entire funnel.

The pause hypothesis

Under 30 minutes
from blank page to
live strategy.

The pilot hypothesis: in a world where AI makes building instant, a structured strategic pause is worth more than the time it costs. The question was whether users would take it, and whether it changed what they built.

The data shows the pause is almost entirely human thinking time. The AI part is nearly instant. The real cost is 28 minutes of structured input. What follows takes under 2 minutes.

~28 min

Human brief input

The thinking work. Goals, assumptions, constraints, design challenge. Measured from one session. Preliminary data point.

56 sec

Brief generation

AI deconstructs the brief into structured strategy components. Goals, risks, assumptions, positioning. Measured from production.

52 sec

Strategy + hypotheses

Competitive positioning, design hypotheses, decisions framework, action plan. Full strategy layer generated and ready.

<30 min

Total: blank to live strategy

From zero to structured strategy with context, hypotheses, decisions and MCP context ready for AI builders.

Traditional equivalent

2-4 weeks

Discovery workshops, stakeholder interviews, competitive research, strategy documentation. Before a line of code is written.

Was the pause worth it?

For builders, yes.

Users actively building when they used Pabs showed the highest workflow fit, highest ratings, and the strongest payment intent. The pause paid off when there was something to build against.

Token economics

7.16 million tokens.
86% went to strategic conversations.

Where the tokens went revealed what to optimise next.

86.4% of tokens were agentic labour.

That is not overhead. That is the product working. 6.2 million tokens of AI sitting alongside a user, reading their live strategy, responding to decisions in real time, adjusting direction on-demand. The conversation layer is where the strategy stays alive between generations. It is the always-on strategic support that used to require a person in the room.

Generating a strategy costs almost nothing.

The actual generation events (brief, hypotheses, competitive analysis, decisions) accounted for less than 1% of all token spend. Creating structured strategy is extraordinarily cheap. The cost is in keeping it alive: the conversations, the pulse check-ins, the learning evaluations running alongside an active build. That is where the value is. That is also where the tokens go.

Model selection

Not one model.
The right model
per job.

Model selection was itself a strategic decision. Each task type has a different performance requirement and a different cost ceiling. Matching them deliberately kept output quality high and token spend low.

ModelCallsTokensCost (USD)Why this model
gpt-5.3-chat-latest2887,742,506$14.68Conversation: highest reasoning quality needed
gpt-5.4106852,118$2.73Strategy, Pulse, Learning eval: structured analytical depth
gpt-5.4-mini889,206$0.07Lightweight Pulse: consistent, lower cost per nudge
gpt-4o1878,920$0.25MCP queries: stable API, well-tested for tool calls
gpt-4o-mini61,050<$0.01Actions: fast, cheap, right for simple task execution

Conversation needs reasoning and memory. Strategy generation needs structured analytical depth. Pulse needs consistency. Actions need speed. Treating every task with the same model would have tripled cost with no quality gain.

Where tokens went

Every token
accounted for.

Conversations
86.4%
Pulse
9.5%
Learning evaluations
2.4%
MCP queries
1.1%
Strategy generation
<1%

The strategic thinking users came for is less than 1% of actual token spend. The real cost is conversation context overhead. Knowing this changed how the product loads and prioritises AI calls.

Credit economics

The cost of a living strategy.

We tracked the AI token cost of every strategic action across the 3-month pilot: 76 strategies, 62 market analyses, 102 conversations, 40 learnings evaluated, 116 scheduled reviews and 70 agent tasks. 466 outputs in total. Total AI token cost: $17.73. That is $0.04 per output. This is what it cost Pabs in tokens to serve, not what users paid.

Getting from a rough idea to a structured strategy with market analysis, decisions and an action plan cost $0.31. From there, scheduled reviews ran every week and observations from the build fed back in for evaluation. Each action had a measurable cost.

A month of keeping one active project alive ran about $0.20. The first month including setup came to around $0.51.

A freelance product strategist on a monthly retainer to do the equivalent work runs $3,000-5,000 per month.

$0.31

Full setup cost

Brief, strategy, market analysis, action plan and first scheduled review. Strategy live and watching.

$0.03

Per scheduled review

Pabs checks what has changed in the build and evaluates whether the strategy still holds.

$0.02

Per observation evaluated

An observation from the build is mapped against the live strategy. Impact assessed. Human decides what changes.

$3-5k

Monthly freelance equivalent

A product strategist on a monthly retainer doing the same ongoing work. Sourced from 2026 market rate benchmarks.

Users access Pabs through a credit system. Credits are priced based on these token costs plus operating margin. Each action type has a defined credit cost. The first complete setup uses around 16 credits. Keeping one active project alive runs about 10 credits a month. The free tier includes 30 credits.

The trust signal

70% of AI-proposed
strategy changes
were trusted.

40 learnings reached the strategy. Pabs evaluated each one, identified what it affected, and proposed what should change. Humans reviewed every proposal. 28 were approved and written back. That 70% approval rate is a trust signal in both directions: humans trusting the AI's judgment, and the AI correctly reading what was critical enough to act on.

How learnings arrived

24
14
2
Detected from chat
Via MCP
Manually submitted

The system surfaced most of them. Nobody had to remember to submit.

What humans decided

28
8
4
Approved · 70%
Pending · 20%
Rejected · 10%

The 10% rejection rate proves the human was genuinely reviewing, not rubber-stamping.

What changed in the product because of those learnings

Activation flow rebuilt

The drop-off before first strategy generation was the sharpest cliff. Onboarding was redesigned around getting users to that first moment faster.

Selective context loading

Token data revealed the chat model was loading the full project context on every turn. Context loading became targeted, not total. Immediate cost and performance improvement.

Multi-plan workflows prioritised

Consultants asked about running multiple client strategies immediately. The first commercial signal shaped the next build cycle.

What the pilot changed

Five things
the data moved.

01

Activation is the bottleneck. And the who matters as much as the when.

Users who generated strategy at least once tended to return. But the stronger signal was about segment: the users who found the most value were actively building when they arrived. They connected Pabs strategy to their build tools via MCP, kept the strategy alive across iterations, and felt the feedback loop working. Pre-build explorers and post-build reflectors saw it as a tool. In-build operators experienced it as a system.

02

The users who would pay were the ones with a live build

All eight users who said they would pay were actively building a product during the pilot. Not exploring ideas. Not reviewing past work. Building now. The willingness to pay wasn't about the tool's features. It was about whether the strategy was connected to something moving. That reframed the pricing question entirely.

03

Consultants asked about clients immediately

The first commercial signal came from people thinking across multiple client strategies simultaneously. Multi-plan workflows became the next build priority.

04

Depth lives with champions, not the crowd

Six champion users accounted for most depth, feedback and product learning. Pulse, learnings and agent tasks weren't discovered organically. Getting users there became the next focus.

05

The product is the research

The pilot wasn't just user research. It was research into whether a strategic AI layer could build, operate, and learn from a live product without a traditional team. The product itself was the experiment.

The number
$17.73

Total AI API cost across the entire three-month pilot. 7.16 million tokens. 46 users. 76 strategies generated. 62 market analyses. 102 conversations. 116 pulse check-ins. 70 agent tasks. $17.73 all in.

All participant names are pseudonymised with permission. Token counts and costs are drawn directly from production database records. Learning counts from plan_learnings table. Agent task counts from agent_tasks table. Pulse counts from plan_pulses table. Data sourced May 2026.

The data conclusion

One segment.
Solo Founders with
something building.

Highest workflow fit

Solo Founders scored 4–5 out of 5. The segment with the clearest product-to-workflow connection.

Strongest pay signal

6 of 8 would-pay users were Solo Founders. All had an active build in progress when they started.

So how much is it worth?

$0.31 to start. Then $0.03 per pulse. $0.02 per learning. The ongoing cost scales with how alive the strategy is. Not with overhead.

Consultants showed early commercial instinct: they asked about client use immediately. But they didn't yet have a workflow that Pabs could anchor to. Innovation and Growth segments showed almost no conversion signal. The data doesn't suggest chasing them next.

The pattern was consistent: value followed workflow fit, workflow fit followed an active build. The product doesn't help people think about building. It helps people who are already building think more clearly.

The model behind the data

See how the
product actually
works.

Pablo Dunovits

Research published May 2026 · pabs.co.nz

Try Pabs free pabs.co.nz