Independent diagnostics for engineering organizations

You spent a year's budget on AI tooling. What did it change?

I measure it, and tell you what to do next: where AI helped your engineering workflow, where it cost you, and the one thing to fix first. Fixed price, two to three weeks.

Book a scoping conversation (30 min) See exactly what you get

The problem

Why the answer isn't already on a dashboard

Google's 2025 DORA report put AI adoption among software development professionals at 90%, and found 30% trust AI "a little" or "not at all." Adoption is settled. What it did to your engineering workflow is not — and three things stand between you and that answer.

01

Usage is not outcomes

Your AI tools report seats, sessions and suggestions accepted. That measures the tool being used. It does not tell you whether the work got faster, slower, safer or more fragile — and no vendor dashboard is built to.

02

Outside research can't answer it for you

Measuring AI's effect on software work across a whole population falls apart once you can't control who takes part and which tasks they hand over. People opt in, and the work where they expect the most help is the least likely to end up in the comparison. That's a limit of the method, not a failing of the researchers.

03

Inside one organization, it's tractable

Nobody is a volunteer. Everyone is already being paid to do the work. Telemetry can be paired with candid interviews instead of self-reported surveys, and everything else that changed alongside AI can be written down rather than assumed away.

The work

One fixed-price engagement

Named, priced and scoped in public, so you know what you would be buying before we ever speak.

AI Workflow Triage

The question it answers

We spend $X a year on AI tooling across our workflow. Where did it help, where did it cost us, and which one thing should we fix first?

What it looks at

Four areas of the engineering workflow: coding, code review, testing and in-repo documentation.

What I look at, in your systems

  • Twelve months of version-control history (GitHub/GitLab) and project-tracker data (Jira/Linear)
  • Your pull-request comments
  • What you pay for AI tools, and who has seats
  • Six to ten interviews
  • A short adoption survey

How the comparison works

A measurement like this is won or lost here, so this part is stated openly rather than left as method.

  • Teams that adopted at different times are compared against each other, rather than the whole organization against a single before-and-after date.
  • Engineers hired after adoption define what ramp time means now. They have no pre-AI baseline, and pretending otherwise is where most of these numbers go wrong.
  • Hiring freezes, layoffs and roadmap changes are written into the report as a stated list, not quietly excluded.
  • Team-level only. No individual scoring, ever. The team is the right unit of measurement — and it is also why nobody on your team has to treat this as surveillance.

What arrives at the end

  • A per-area scorecard: helped, neutral, or cost you
  • A paired read — throughput and stability and what it cost in attention. Never throughput alone.
  • A spend map
  • One named bottleneck, with three to five ranked moves
  • A one-pager written for your board
  • A follow-on already scoped, if you want one

Price and time

$12,000 – $20,000 Fixed, agreed before work starts. Two to three weeks.

What it is not

  • Not a tool bake-off
  • Not a platform to install
  • Not performance reviews
  • Not an exhaustive measurement of everything — it names one thing to fix first

I don't sell you the tools, I don't install a platform, and I'm paid the same whatever the data says.

Book a scoping conversation (30 min)

Fit

Who this is for

Saying this plainly saves us both a call.

A fit

This is for you if

  • You run engineering at a B2B software product company — SaaS, developer tools, AI, marketplaces
  • You have roughly 30 to 90 engineers
  • AI tooling is a budget line, not a few seats
  • You are the VP Engineering or CTO who owns that line
Not a fit

This is not for

  • Agencies and consultancies
  • IT departments of companies that don't ship software
  • Anyone looking for help choosing, buying or rolling out AI tools — that is the opposite of what this is

Research

Conversations with engineering leaders

I'm running a series of conversations with engineering leaders about what AI has changed on their teams — what they've measured, what they haven't, and where it has helped or hurt. Fifteen minutes. In exchange you get the aggregate results across every conversation.

This one is not a sales call. If you would rather talk about the Triage, book the scoping conversation instead.

Book 15 minutes
Kwaku Farkye, founder of Afram Intelligence

About

From first engineer to enterprise scale

I'm Kwaku Farkye. I've spent over ten years building software — as a founder or founding engineer at five early-stage startups, two of which were acquired, and as a technical lead and CTO. I was the first full-time engineer at Rimeto, which Slack acquired. I'm based in San Diego.

Before this I was co-founder and CTO at Navi, which ran innovation programs for organizations across the US and wound down operations in December 2025. Navi was a Department of Defense innovation contractor from 2021 to 2025 and also ran programs with universities. Over its life it delivered 49 programs for 31 organizations and trained more than 1,200 people; teams that came through those programs have raised $243.9M between them.

Every one of those programs ended with a report on what actually worked. Writing those honestly — separating what the program did from what would have happened anyway — is much harder than it looks, and it is what this practice is built on.

Afram Intelligence is my practice, and I am the only person in it. Being independent isn't a positioning choice here. It is the only reason a read like this is worth paying for.

Navi → LinkedIn →

Contact

Two ways to start

They are different meetings. Pick the one that matches why you are here — a conversation booked under the wrong premise wastes your time and spoils mine.

If you're buying

A scoping conversation

Thirty minutes on your engineering organization, your AI spend, and whether the Triage is the right thing to buy. You will get a straight answer if it isn't.

Book a scoping conversation (30 min)
If you're curious

Fifteen minutes for the research

No pitch. I ask what AI changed on your team, and you get the aggregate results across every conversation.

Book 15 minutes

Or email me directly: kwaku@aframintelligence.com

Prefer to write?

This reaches the same inbox.

I'll respond within 1–2 business days.