We build the systems we advise on.
Strategy and applied AI engineering from the same practice — so the advice is answerable for the build, and the build is designed to survive handover.
Executive outcomes, AI-powered.
Selected work
Three examples of advice made operational.
Real data, running systems, explicit limits — and a handoff designed so the client can keep moving without us.
SME & association operations
Smriti
A residential community's committee now asks questions of its own records in WhatsApp — and gets answers, from an agent that cannot change those records.
3 independent write-blocks · live daily use
Physically unable to write
The query path opens the database read-only. Not instructed not to edit — unable to. Verified against the live database, including the negative tests.
Enterprise CXO visibility
CTO Dashboard
A demo the client extended himself — onto his own SharePoint, across a 119-project portfolio — and now runs without us.
119-project live portfolio · client-operated
Which statuses are inferred
Blocked and paused are explicit. At-risk is inferred from free text, and the dashboard says so before a CXO takes it into a board review.
SME finance & receivables
The Collections Agent
A daily receivables report assembled from the sales team's own chat group and emailed each evening — a number the business used to get monthly, if at all.
Monthly → daily receivables visibility
The numbers come from a parser
Code extracts the figures; the model only writes the commentary around them. It never does the arithmetic.
The recurring question is simple: what prevents a useful AI system from becoming an unbounded one? We prefer answers enforced in code, not promised in prompts.
"The use of dialogue from movies is a very unique touch — don't recollect seeing it in this depth before… the book makes for a breezy read and a fun addition to your library."
Dr. Sai Praveen HaranathApollo Hospitals, Jubilee Hills, Hyderabad
What you can engage us for
Four ways we typically help.
Engagements are shaped to the problem, not a package. Most combine advisory with a working implementation or a decision-ready operating artifact.
Agent systems for operations
A bounded agent on your real data, in the channel your people already use, with the write path controlled at the database rather than in the prompt.
Proof: Smriti
Executive visibility
Dashboards a CXO actually opens — that distinguish what's known from what's inferred, and that your team can extend without calling us.
Proof: CTO Dashboard
Creative & content operations
For teams whose AI experiments produce plausible, unusable output. Map the real process, encode the stages as gates, stop the model skipping ahead — then automate planning, production and publishing.
Proof: AdCraftAI, Movisvami
Advisory, made observable
Financial health and fiscal architecture for SMEs; executive positioning and second-act planning for CXOs. Uncomfortable truths, delivered with structure — and where it helps, the numbers automated so they arrive daily rather than quarterly.
Proof: The Collections Agent · Led by the principal
How we work
The same four steps, every time.
A short loop: understand the real problem, put something real in front of it, harden what matters, then make the result transferable.
Start from the real problem
What's actually being asked, and of what data. Almost always smaller and more specific than the opening brief.
Ship something running
On your real data, in days. Real enough to argue with — because a demo nobody disagrees with taught you nothing.
Harden it
Write paths, identity, authorisation, and the honest declaration of limits. The step most AI pilots skip, and the reason most of them stall.
Make it repeatable
Export scripts, schemas, handoff prompts, checks. Success means the next iteration doesn't need us.
From the principal
Why I still write the code.
I spent twenty-five years running operations before I wrote a line of this. COO of a digital cinema company. Chief-Digital at a broadcast group. Enough board reviews to know what a dashboard is really for, and enough failed rollouts to know why most of them don't survive the second quarter.
I don't think you can advise credibly on systems you've never had to make work on a Tuesday night.
So I build them. Smriti runs in a real building, on real money, and I hardened its database handle myself after deciding a prompt instruction wasn't a control. That's not a marketing story — it's the reason I'll tell you honestly when something isn't worth building, and why that answer is worth what you'd pay for the opposite one.
The firm has grown past me, which is the point. But the judgment you're buying at the start of an engagement is mine, and I'd rather you knew whose it was.
The practice takes its name from Directing Business — the 2022 book that used movies as a way to think about management, judgment and execution.
Lakshmi Narayana U
Founder & Principal Consultant · Directing Business Consulting & Advisory
Background
- COO — Cinematica Digitals, digital cinema subsidiary of Geetha Arts
- Chief-Digital — Rachana Television (NTV, Bhakthi TV, Vanitha TV)
- Tech Mahindra — earlier career
- 25+ years across internet, media and entertainment, with an entrepreneurial streak kept intact throughout
- 243+ essays published on LLMs, agents and the business of AI
Directing Business (2022)
Management lessons from extraordinary movies around the globe — the book that named the practice, and explains why every system here gets read like a screenplay first.
Writing & research
We publish the working out, not just the conclusions.
Over 243 hands-on essays on LLMs, agents, RAG and the business of AI — most written while the thing was still breaking.
Medium
Long-form essays on AI practice, model releases tested rather than summarised, and the occasional post-mortem of a build that didn't work.
AI Insights
An AI practice companion for LLMs, RAG and agents — grounded in the full essay archive. Architectures, trade-offs, hands-on guidance.
Director's Cut
The weekly digest — what shipped in AI, notable reading, what we built, and the film that framed it. One email, one sitting.
"I am tired of reading management books which are prescriptive and seem to know answers to all the problems. This book introduces management concepts in a way that is unique and compelling… I loved the way the author correlated management concepts to a scene in a movie."
Sriram UCXO
Start a conversation
Tell us what's actually breaking.
Not what you think the AI solution is — that part is usually wrong, including when we guess it. Describe the problem and you'll get an honest read on whether it's worth building.
First conversations are with Lakshmi Narayana U, Founder & Principal Consultant.