Book a 20-minute callBook a call

One-time engagement

Decision session

A 90-minute session on one decision in your AI stack, a vector database, an agent framework, a model provider or build versus buy. You bring the shortlist; you leave with a decision you can defend, written up within two days.

One decision is holding up the quarter. Make it in 90 minutes.

  • 90 minutes
  • $499
  • The decision and the reasoning, written up
Book a 20-minute call

Is this for you?

  • One decision is blocking the quarter

    The team has a shortlist, opinions on both sides and no way to settle it.

  • The vendors all sound right

    Every demo looked good, and none of them showed you what it costs to leave.

  • Build or buy is the question

    You could build it in a quarter or buy it today, and the real cost of each is not on anyone’s slide.

Not this? An Hour with the Architect is the 60-minute form for one hard problem worked live ($129); the architecture review is for a whole system about to scale.

What you get

  1. 01

    The decision

    Which way to go, stated in one line, with what the decision turned on.

  2. 02

    The reasoning

    Each option against the five things that matter for you, so you can defend the call to your team and your board.

  3. 03

    The cost of changing your mind

    What it takes to switch later, so the decision is made with its exit in view.

Illustrative pages. Yours carry your system’s scores, numbers and fixes.

What it costs to reverse

The reasoning

  1. 1CostFix
  2. 2CostFix
  3. 3CostFix
  4. 4CostFix
  5. 5CostFix

The decision

One choice, five things it turns on

  • What it has to do next year
  • Volume and latency
  • What it costs to run
  • What it costs to leave
  • Fit with your team

What the decision turns on

Five things every stack decision turns on. We put your shortlist against each one, live, with your numbers.

  1. 01What it has to do next year

    I take what you need it to do in the next twelve months, not the demo, and test each option against that.

    I measureEach option against next year’s needs

  2. 02Volume and latency

    I take your real volumes and the latency you have to meet, and find which option still works at ten times.

    I measureEach option at today’s volume and at ten times

  3. 03What it costs to run

    I plot what each option costs at your volume, with the people it takes as well as the bill.

    I measureThe cost curve for each option

  4. 04What it costs to leave

    I weigh how hard each option is to leave, the data, the integrations, the lock-in, so the decision keeps its exit.

    I measureThe cost of reversing each option

  5. 05Fit with your team

    I check what each option asks of the people you have: the skills, the on-call, the single person who would end up owning it.

    I measureWhat each option asks of the team

90 minutes, the write-up within 2 days

How it runs

  1. Before

    Send the shortlist

    The options, what you need them to do in the next twelve months, your rough volumes and any constraint that forces the choice.

  2. The session

    Work it through

    Ninety minutes, live: each option against the five things it turns on, with your numbers, until the decision is clear.

  3. Within 2 days

    The write-up

    The decision, the reasoning and the cost of reversing it, written so you can defend it to your team.

The next step

Start with a
20-minute call.

Tell me the decision and the shortlist. If the session fits, you book it on the call; if it needs more than 90 minutes, I say so.

Bring

  • The decision, in a sentence
  • The shortlist as it stands
  • The date it has to be made by
Book a 20-minute call Audio, at a time you pick. No deck, no preparation.

Why me

I have made these calls on systems that then had to live with them. The stacks at Devnagri AI and TWINii run on Node.js and TypeScript, Python, PostgreSQL, MongoDB, Redis and the major clouds, chosen for what each system had to do rather than for what was in fashion. I hold no vendor relationships and take no referral fees, so the answer has no side.

The track record
“It's like he can see inside your head what you want to explain and will start improvising on them, and then all you will be saying is yes, yes, exactly.”
Kumar RaviAssociate Staff Engineer, Nagarro

Asked on most first calls

The questions.

What should I prepare?

Your shortlist, what you need it to do in the next twelve months, your rough volumes and any constraint that forces the choice: budget, data residency, team skills. Send it when you book so the 90 minutes go on the decision.

What if we have not narrowed it down yet?

Send what you have. Part of the session is cutting the list to the options that can actually do the job.

Do you have vendor relationships?

No. No vendor relationships, no referral fees, no equity, so the answer has no side.

What if the answer is to build it?

Then the write-up says what to build, what it takes and what you give up, the same as for any other option.

Can I share code or architecture on the call?

Yes. I sign your NDA first if you need one.

What do I get afterwards?

The decision, the reasoning and the cost of reversing it, in writing, within two days. It is written so you can hand it to your team or your board as it is.

What if the decision needs more than 90 minutes?

I say so on the first call, and we talk about the architecture review or the demo-to-production plan instead.

Every way I help

Loading the calendar.