Six Week AI
Ship an AI workflow in six weeks. Train your team to own it.
I help mid-size companies pick the one internal workflow where AI delivers measurable ROI right now — then I build it with your team. Six weeks. Fixed fee. One working AI workflow shipped with its impact quantified in hours and dollars saved, plus a roadmap for what comes next.
Start here · free
You know AI should be doing more in your business. You can’t tell which workflow to point it at.
So the projects stall. A pilot gets built, demos well, and quietly dies — because it was the wrong workflow to begin with. Meanwhile the pressure to “do something with AI” only goes up.
Every workflow looks like an AI candidate — and most aren’t. Get a way to tell the difference before you spend a quarter and a budget finding out the hard way.
“AI makes this newly possible” vs. “we could’ve scripted this years ago” — most teams can’t tell them apart. The test scores both, so you back the workflows where AI is the actual unlock, not the ones that just sound impressive.
“Will it be accurate enough?” keeps killing momentum. Get the lenses that tell you up front whether a workflow can clear its own accuracy bar — or whether it’ll always need a human in the loop, and what that costs.
Go from “we should be doing more with AI” to a clear, repeatable way to score any workflow in your company as a good bet or a money pit — in an afternoon, not a quarter.
Already know you want to talk? Book a call · hello@ajitkrishna.com
Who this is for
You need to move on AI.
Nobody can tell you what that actually means.
You run a company — or a department inside one — with somewhere between 200 and 2,000 people. You know AI has to become part of how you operate. What you don’t have is a clear, credible answer for what that looks like in your business, this quarter.
The fit window
Where the engagement actually lands.
Company size
200–2,000 people
AI maturity
Real workflows, no platform team yet
What does “AI-native” even mean?
The term is abstract — everyone defines it differently, and no one’s sure where to actually start.
Consultants keep it strategic
You’d walk away with a deck, not something that works.
Hiring is a slow, expensive gamble
And you can’t easily tell who knows the frontier from who’s bluffing.
You can’t tell which approach actually works
And the wrong bet is expensive and slow to unwind.
I call this engagement Six Week AI: a six-week, fixed-fee build that picks one internal workflow, ships it with your team, and leaves them able to own the next one.
If you’re earlier (no real workflows yet) or much larger (already running a mature AI platform team), I’m probably not the right fit — and I’ll say so on the first call.
What AI-native looks like
Abstract gets concrete fast.
Three workflows AI makes newly possible — the shape of what we’d build, by department.
Legal
Contract intake
Inbound contract → key terms, dates, and red flags extracted to a structured record.
Finance / AP
Invoice processing & GL coding
Invoices in → line items matched, GL codes suggested, anomalies flagged.
Customer Success
Renewal-risk briefing
Account signals in → a one-page pre-renewal briefing out.
None of this is “paste it into a chatbot.” Each one is a designed system — and that engineering is what separates a demo from something a business can actually run on.
The deal
I take the risk out — and you ship something real.
It starts with a conversation, not a contract.
A consultative call to see if there’s even a fit. Sometimes there isn’t, and I’ll tell you.
Mapping comes before building.
I map how your team actually works and show you where AI pays off — and why an AI-first approach makes sense — before you commit to a build.
I build it, your engineering counterpart builds with me.
We write the AI workflow and the integration code together, against your real systems and your real data. By week 6 your engineering counterpart has co-authored the codebase and owns the deploy path.
Your team owns it.
I train them and hand it off, so the next workflow doesn’t need me — unless you want me back for it.
Ownership transfer
Where the codebase and the deploy path end up.
At the end of six weeks, you have
- A working AI feature against the chosen workflow, running on your data, integrated where it needs to be
- A written workflow map of how the department operates today
- A prioritized roadmap of the next 3–8 AI opportunities in that department
- A playbook: how the workflow works, the key decisions, and the eval criteria with how it performs
- A trained internal team capable of taking the next workflow on without me
Total time from your team: about 8–12 hours over six weeks for stakeholder interviews, prototype reviews, and the workshop — plus the dedicated time of your engineering counterpart.
How it works
The six weeks
Pick a department — legal, finance, customer success, sales ops, support, HR. Over six weeks I do five things:
Step 1
Learn the workflow
I sit with the people doing the work — not the org-chart version — and map where information enters, where it gets stuck, and where humans add real judgment.
Step 2
Find the payoff
I score every opportunity on value and on AI-fit. We build the one at the intersection — high-value, and solvable only because of what these models can now do.
Step 3
Build it together
Hands on the keyboard, paired with your engineer. We write the AI workflow and integration code on your real data — built to be evaluated, not a demo that hallucinates in production.
Step 4
Leave a playbook
How the workflow runs, the key decisions, the eval criteria and how it performs, and the next opportunities worth pursuing. Take the next one yourself, or with me.
Step 5
Train the team
A workshop with whoever runs the workflow. They leave able to spot the next AI opportunity themselves — which is how an org turns AI-first, one team at a time.
What this is not
Clear about the boundaries.
Not a fractional Head of Product
Async, with two scheduled review checkpoints during the build.
Not an AI strategy report
A shipped feature in production — and a team that thinks AI-native enough to find the next ones.
Not a longer engagement in disguise
One workflow, six weeks. Want a second? Separate engagement, same price.
Why me
Twenty years across product, design, and engineering.
Most recently at
Product leader on the platform team behind its workflow engine.
Consumer-scale product design, to millions of users.
Service-design rigor, at national scale.
The ladder
The whole path, with prices.
No mystery about cost. Here’s the whole path, from a free self-check to a shipped workflow. Most people start at the top and move down only as far as they need.
Tier 01
Free
Find out if your workflow is an AI fit
- The AI Workflow Litmus Test — a 1-page set of evaluation lenses
- A short personal video teardown of your specific workflow
Tier 02 · Most start here
$2,500
Credits in full toward a build.
Workflow AI Mini-Workshop
- 90-minute working session, live with you and your team
- Map the workflow, score AI opportunities, prioritize, assess risk
- You leave with an AI-redesigned workflow and a rough ROI
Tier 03
$25,000
Fixed fee · 50% kickoff / 50% handover.
The Six-Week Build
- One production AI workflow, built with your engineer on your real data
- Codebase handed off; your team owns the deploy path
- Team trained so the next workflow doesn’t need me
The build is $25,000, fixed fee, one workflow. The mini-workshop is $2,500 and credits toward it. The Litmus Test and a personal teardown are free.
Frequently asked questions
The questions buyers actually ask.
What if we want you to actually ship to production, not just hand off a prototype?
The default engagement is production-ready. By week 6 your engineering counterpart has been writing integration code with me throughout the build, and the artifact lives in your stack. What’s not in scope is operating the system long-term (monitoring, on-call, evolving prompts and evals as the model changes, retraining on new data). That’s a separate engagement — a 90-day “embed” retainer, scoped from what we learn during the six weeks. Most clients run the system themselves; some come back for a retainer once they see the maintenance shape.
Which department should we pick?
We can decide together on the first call. Departments where I’ve seen the strongest AI-workflow opportunities recently: legal (contract intake, redlining, NDA review), finance (invoice processing, AP automation, expense workflows, close acceleration), customer success (renewal preparation, account health synthesis, QBR prep), customer support (deflection, agent assist), sales operations (lead qualification, CRM enrichment, proposal drafting), HR and recruiting (resume screening, interview scheduling, employee help-desk). If you have a specific department in mind, we’ll talk about whether it’s a good fit. If you don’t, I’ll help you pick.
What AI models and tools do you work with?
Primarily Claude and the OpenAI frontier models. Cursor for hands-on building. I’m model-agnostic in principle — I use whatever fits your data, latency, cost, and privacy constraints best. If you’re locked into a specific cloud or vendor (Azure OpenAI, AWS Bedrock, Vertex AI), I work within those bounds.
How do you handle our data and IP?
Standard mutual NDA before discovery starts. I don’t train models on your data, I don’t reuse your data across clients, and the feature + all deliverables are yours. If you have specific compliance constraints (SOC 2, HIPAA, GDPR data residency), raise them on the first call and we’ll work out what’s possible within the scope.
What do you need from our engineering team?
Your engineering counterpart for the build — someone senior enough to make integration decisions in your stack, with time freed up across weeks 2–6. This is a genuine collaboration, not grunt work: we pair on the build and write the AI workflow and the integration code together, and they go as deep into the AI engineering as they want to. By week 6 they co-own the codebase and the deploy path. They don’t need prior LLM experience — teaching your team to build this way is part of the point. We’ll define exactly what’s needed in the proposal after the first call.
Can the same engagement cover a second department?
No. The six-week scope is one department. If you want AI work across multiple departments, we do them as separate engagements, sequentially or in parallel. Many clients end up doing two or three departments over six months.
Can the engagement be shorter or longer than six weeks?
Six weeks is the box. Long enough to do real work, short enough to keep momentum. I don’t sell three-month engagements because they invite scope creep; I don’t sell two-week sprints because they don’t produce shippable work.
Why fixed fee instead of hourly?
I get faster every time I do this, and hourly punishes that. Fixed fee aligns incentives: you get a known outcome at a known price, I get predictable revenue, neither of us watches the clock.
Do you offer team training as a separate engagement?
Yes. After an engagement, many clients want a deeper team training day — a structured workshop for their PM and engineering leadership on how to scope, build, and manage AI features going forward. I offer this as a separate engagement, priced based on team size and depth. Ask on the call.
Where are you based, and where will you work?
The San Francisco Bay Area. The engagement runs async-first over video. I don’t fly to client offices unless there’s a specific reason, and if there is, travel is billed separately at cost.
How quickly can you start?
Usually two to four weeks out. I run one to two engagements concurrently, so availability rotates. The first call is the right place to check current availability.
If that sounds like your situation, start here.
Run a workflow through the Litmus Test. You’ll know whether it’s the right one to bet on — and what comes next if it is.
Already know you want to talk? Book a call · hello@ajitkrishna.com