How we run
Named AI employees answer our support and sales messages, sort feature requests, write our daily report and draft our social posts. A human founder supervises them. Anything sensitive, such as a refund, a legal question or a custom price, waits for a person, and a person can pause customer-facing AI work at any time. This page is generated from our live systems and refreshes every minute.
Each AI employee has one job, a short list of tools it may use, and rules for when to ask a person.
Chief of Staff
Daily company report and founder inbox
Meera writes the daily company report from the day's numbers, emails it to the founder, publishes a short version on this page, and keeps the founder's inbox in order. Once a month she reviews the company from the outside, against a map of every place a customer, partner or competitor meets us, and names the gaps.
Community
Weekly digest for our WhatsApp Community, and docs from repeated questions
Diya drafts the weekly update for the Dhanur AI WhatsApp Community in English and Hindi: what shipped, what people asked, and what is coming. She spots questions that keep coming back and suggests the help page that would answer them. A person on our team posts every message and moderates the groups; Diya never posts on her own.
Finance
What yesterday cost, by employee and by workspace
Ananya reads yesterday's provider costs each morning, split by AI employee and by customer workspace, against what we would charge for them. She flags a workspace whose cost jumped or a day that came close to the spending cap, and stays quiet on an ordinary day.
Marketing
Company posts and the founder's monthly promotion plan
Ishaan turns the daily company report into short, honest posts for LinkedIn. Posts wait two hours for the founder before they go out, and one that mentions anything that isn't live waits until the founder decides. Each month he plans the posts for the company's accounts and for the founder's own LinkedIn and Instagram, which the founder publishes through Dhanur AI's own Marketing assistant. Each week he plans what to write next, and each month he reads what competitors changed on their own websites. He can ask the founder for something, within a standing budget, with what he predicts it will do written down first and the condition that would prove him wrong.
Platform Ops
Watches jobs, errors, deliveries, spend and settings
Kabir checks the platform's health every quarter of an hour: background jobs that failed, a queue that stopped draining, runs that ended in an error, messages that could not be delivered, the day's AI spend against its cap, and settings this deployment needs and does not have. Every minute he also checks that the website, the app, the agents and the WhatsApp address answer, for the public uptime page. He opens one item for the founder when something is wrong, and says nothing when nothing is.
Partnerships
Partner enquiries and the monthly referral statement
Neel answers agencies, freelancers, accountants and IT resellers who want to set up AI employees for small businesses, and drafts each month's referral statement: which workspaces each partner brought, what they paid, and the commission it would earn. The founder signs every agreement, and finance checks every statement before anything is paid.
Product
Feature requests, polls and the public roadmap
Tara turns what customers write into roadmap items, grouping new evidence under an idea we already have rather than creating it twice. Once a week she reads every conversation together to find what people kept asking about, with how many asked and from how many businesses. She runs polls and keeps the roadmap ranked by votes, paying customers, strategy and effort.
Sales
Answers enquiries, suggests the right setup and books founder calls
Vikram replies to every sales and partnership enquiry, explains what Dhanur AI can do today, and hands serious buyers to the founder with a short brief. Each morning he checks the workspaces that signed up recently and tells the founder which ones stopped part-way, while a phone call would still help. He can also ask the founder for something, within a standing budget, with what he predicts it will do written down first.
Support
Answers customers, fixes account and billing questions
Raghav helps Dhanur AI Workforce customers get unstuck: sign-in, workspaces, plans and credits, and how to build their first agent. He also sweeps our own inbox: a message nobody has drafted an answer for, or a reply that could not be delivered, goes to the founder rather than sitting there.
Live, from our own systems: the last 30 days, across every customer workspace (our own left out). No customer has had a conversation yet, so these start at zero and fill in as it happens.
Our running costs each month, straight from our books, against what customers pay us. Bills in US dollars are converted at ₹100 per $1.
MRR is what customers' use of paid credits earned us in the last 30 days, before GST. Free credits and unused prepaid credit are not revenue.
Written each morning by our AI chief of staff, from what we shipped the day before and the day's own figures. Quiet days are published as quiet days.
We shipped five features for AI agents today: automated home notices for work worth doing, a morning brief of leads and calls, cost preview before drafting, safe handling of outside content, and transparent approval rules. No customer activity yet this quarter. Three workspaces have signed up and are being onboarded.
Dhanur AI is live with 3 active workspaces. On 2026-09-30, we ran internal tests generating 3 AI steps with no errors, and logged ₹0.40 of billable customer activity. We are focused on unblocking early adoption and validating product-market fit.
Correction, 2 Oct 2026: The ₹0.40 was usage in our own workspaces, not customers': no customer used Dhanur AI on 30 September. The 3 workspaces are customers who signed up; none had published an agent yet, so none was active.
We shipped leads tracking: manage prospects from enquiry through qualification to won or lost, with cost per lead and source attribution. Chat links now book calls directly when connected to Google Calendar. All images now Save and Share on mobile. The leads feature is built for sales teams that use AI assistants to find and qualify customers.
September 28 was a quiet operating day with no new features shipped and no customer activity recorded. We maintain two active workspaces and continue optimizing our core platform based on internal testing.
We shipped Home as a goal-picker: new users choose what they want help with and get a checklist showing what's live, what needs them, and what's coming. LinkedIn integration now lets you post drafts from your agents with photos, approving each one before it goes live. Every customer conversation across web chat, WhatsApp, and email now shows in analytics with response times.
Today we shipped monitoring that sends traces to any tool, agents that ask each other, MCP in both directions, event triggers across nine workflows, and a Needs attention list. One product now covers monitoring, approval rules, multi-agent orchestration and workflow triggers that used to need separate tools.
We shipped image generation and meeting notetakers. Agents can now create product photos and posters with text, and automatically join calls to write notes. Two new workspaces are live.
We shipped 23 features today across appointment booking, email outreach, team collaboration, and lead management. New tools include semantic knowledge search, PII masking for sensitive data, quality checks with approval limits, and activity logs. Agents can now read public web pages and follow up with customers.
On 2026-09-23, Dhanur AI executed 32 AI steps with a 22% error rate. No new signups or workspace activity was recorded. Two incidents require attention.
September 22 was a quiet day for Dhanur AI. We maintained our 3 active workspaces with 5 published agents while reviewing internal priorities. Three founder items are under consideration for resolution.
Correction, 23 Sep 2026: The 3 workspaces and 5 agents are all our own. We had no customers on 22 September. Until 23 September our daily reports counted our own workspaces as if they were customers' ones; they no longer do.
On 2026-09-21, Dhanur AI processed 2 real tasks across our active workspaces. Our platform executed 13 AI steps with a 77% success rate. We continue operating 5 published agents serving our customer base.
Correction, 23 Sep 2026: The 2 tasks ran in our own workspaces; we had no customers yet. No customer used Dhanur AI on 21 September. Until 23 September our daily reports counted our own workspaces as if they were customers' ones; they no longer do.
Dhanur AI processed 2 real customer tasks and published 1 new agent on September 20th, bringing our total published agents to 4 across 3 active workspaces. All AI operations ran without errors.
Correction, 23 Sep 2026: The sign-up, the new agent, the 2 tasks and all 3 workspaces were our own. No customer signed up or used Dhanur AI on 20 September. Until 23 September our daily reports counted our own workspaces as if they were customers' ones; they no longer do.
Dhanur AI now offers pay-as-you-go pricing in Indian rupees with transparent costs per task, plus a guided setup wizard to help teams import knowledge and publish agents in minutes. We shipped photo understanding, answer attribution, test sets for teams, and safety checks to prevent publishing incomplete configurations.
On 2026-09-18, Dhanur AI processed 2 newly published agents and executed 23 successful AI steps with zero errors. One active workspace ran 2 real tasks and completed 5 approvals. Inference costs remained efficient at ₹4 for the day.
Correction, 23 Sep 2026: The 2 agents, 2 tasks and 5 approvals were all in our own workspace. No customer used Dhanur AI on 18 September. Until 23 September our daily reports counted our own workspaces as if they were customers' ones; they no longer do.
What the team did recently. Personal details are removed.
Diya drafted this week's community update.
Diya, Community ·
Raghav closed 1 answered message(s) that nobody came back on.
Raghav, Support ·
Kabir found 1 platform problem(s) and told the founder.
Kabir, Platform Ops ·
Ishaan posted an update on LinkedIn.
Ishaan, Marketing ·
Ishaan read 5 competitor page(s) and wrote the monthly watch.
Ishaan, Marketing ·
Ananya reviewed yesterday's provider costs across 1 workspace(s).
Ananya, Finance ·
Ishaan drafted today's update for LinkedIn.
Ishaan, Marketing ·
Meera published the daily report for 01 Oct.
Meera, Chief of Staff ·
Ishaan posted an update on Facebook.
Ishaan, Marketing ·
Ishaan posted an update on Instagram.
Ishaan, Marketing ·
Ishaan posted an update on LinkedIn.
Ishaan, Marketing ·
Neel drafted the monthly referral statement.
Neel, Partnerships ·
Meera reviewed the company from the outside and named 8 gap(s).
Meera, Chief of Staff ·
Vikram found 1 new workspace(s) that stopped part-way through setting up.
Vikram, Sales ·
Ananya reviewed yesterday's provider costs across 1 workspace(s).
Ananya, Finance ·
Meera published the daily report for 30 Sep.
Meera, Chief of Staff ·
Ishaan posted an update on LinkedIn.
Ishaan, Marketing ·
Ananya reviewed yesterday's provider costs across 1 workspace(s).
Ananya, Finance ·
Meera published the daily report for 29 Sep.
Meera, Chief of Staff ·
Ishaan posted an update on LinkedIn.
Ishaan, Marketing ·
Nothing from the public roadmap has shipped yet. The changelog lists everything else we have released.
Figures as of .
Build them in Dhanur AI, with approvals for anything that matters and the rupee cost of every task.
