Atif Javed & Daniel Lesliehello@joinridealong.comjoinridealong.com
The problem
Companies bought AI for everyone. Almost nobody uses it.
40M+enterprise AI seats sold
3 in 4go unused in any given week
$600M+a month in seats nobody opens
Public vendor disclosures and independent adoption research, 2026. Waste is unused seats at a blended $20 per seat per month.
It does not need to be cut. People just need to be shown how to use it.
Why every fix has failed
Nobody learns from a slide deck.
Training decks get skimmed
Consultants leave, nothing sticks
Prompt libraries assume you know what to ask, when most people do not know what can be automated at all
Vendor dashboards count logins, not value
Daniel watched his wife spend hours a week on work AI could do in seconds. Her company had done everything right on paper: told everyone to use AI, bought the seats, hired consultants.
He sat down behind her and taught her on her own work. It stuck immediately. It was the first thing that ever had.
She did not need better training. She needed someone in the chair next to her.
Team
Built for exactly this.
Atif JavedCo-founder
Director of Product, Monarch
MIT and YC alum, sold his last startup in 2025
Observability products at Meta and Coinbase
Sold into hospitals, the hardest enterprise buyer there is
Daniel LeslieCo-founder
Staff Engineer, Monarch
Previously Plaid and Atlassian
Three years building in AI: model development at Plaid, financial AI agents at Monarch
Shipped Monarch's MCP server with Atif
A year shipping together at Monarch. RideAlong is observability, pointed at work itself, which is what we have both spent our careers building.
What we built
An ambient desktop coach that teaches people to hand work over.
1 · It spots the busywork
Runs quietly on the desktop, learns the shape of the day, and finds the work that repeats. No prompting, no setup.
2 · It teaches, in the moment
A nudge exactly when the work appears, then a guided walkthrough on the real task. The person does every click, which is why it sticks.
3 · It hands the job over
What was a weekly ritual becomes a routine that runs itself, with a human approving. The skill stays with the employee either way.
It works with the AI your company already pays for. Nothing new to buy, nothing new to learn.
Everyone can count what is automated. We see what should be.
What is automated today. Every tool measures this.Everything that should be and is not. Nobody measures this.
We watch the work itself, at the desktop, across every tool. Process mining sees system logs. Adoption platforms see the apps they instrumented. Vendors see their own product.
The long tail is where the waste lives, and it never touches an ERP log.
We measured the baseline by hand, before automating, so we can price on verified hours saved. No competitor has the counterfactual.
Neutral by necessity. Microsoft will never teach your team Claude. Enterprises run three stacks.
Landscape
Three camps competing for the same budget.
Who
What they do
What they cannot see
WalkMe, Whatfix, Pendo Digital adoption
Generic walkthroughs inside apps they instrumented
Work outside those apps, and whether anything got faster
Celonis, UiPath Process mining and RPA
Find automatable flows in system and ERP logs
The desktop long tail, where most white collar waste sits
Copilot dashboards, agent startups Vendors and agents
Report usage of one tool, or do the work for you
Other vendors' tools, and the skills people never gain
We sell recovery of spend already committed, which makes this a CFO purchase rather than an L&D one.
Business model
We get paid when AI pays off.
Base: $10 to $15 per seat per month, against the roughly $30 per seat companies already spend on AI.
Outcome: $2 to $3 per verified hour saved, capped so finance can budget. About 5% of the value created at a $50 loaded hour.
We can price this way because we measured the baseline before automating.
$60Ka year, one 500 seat department
$600K to $900Ka year, full 5,000 seat rollout
~$290MARR at 2M seats and $12 a month, before the expansion products
Market
Every knowledge worker is about to get a seat. Somebody has to make them pay off.
40M+enterprise AI seats today, growing fast
~$14Ba year in current enterprise AI seat spend
1B+knowledge workers, the long term seat pool
Enablement and measurement is a layer above every vendor, so it grows with total AI spend rather than competing for a slice of it. If AI seats become as universal as email, this is a multi billion dollar layer.
Where we are
One month in.
18customer interviews
90%said they would pay
MVPcapture agent running on our own machines
Native macOS agent shipping semantic capture today, no screen recording and no keylogging
Workflow mining producing real baselines from real work
Coach experience and leadership dashboard live at joinridealong.com
Design partner conversations underway, first cohort forming now
No revenue yet. Both founders go full time on close.
Where it goes
Coaching is the front door. One engine underneath.
NowCoach and measure
Teach people to delegate, prove the hours, show the automation gap.
NextAutomate and spread
Routines that run themselves. Playbooks so one person's skill becomes everyone's.
LaterRun the AI workforce
Automation discovery, spend and vendor scorecards, and supervision for fleets of agents.
Every one of these needs the same thing nobody else has: a live model of how the company actually works.
RideAlong
The work that repeats should not have to.
Bring us one team and one messy week of work. We will show you what RideAlong finds, what it teaches, and what it hands back.