Founder dreamlaunch.studio | dev & design partner for founders and fast-growing startups | worked with bounce, lifesight, mrsam, mizuai + 40 more
Book a call →
- Tweets 9,580
- Following 1,052
- Followers 22,013
- Likes 71,074
enterprise onboarding that looks great in the sales demo:
- SSO, role matrix and a compliance modal before anything else
- a 12 field "tell us about your org" form
- an "invite your team" screen on step 2
- every integration permission asked upfront
- a guided tour narrated by the CSM
- a dashboard that stays empty until IT approves the connector
what we built for tenmo instead:
- SSO is one toggle on step 1, next to a live preview of the workspace
- telemetry sources connect one at a time, each with a read only note right there
- the fleet map fills with drones the moment a source connects
- alert rules replay real flights so the ops lead sees what would have been caught
- inviting the team comes on step 5, after there is something to show them, and it is skippable
- the copilot answers a real fleet question before setup ends
tenmo is a fleet observation platform for enterprise drone delivery.
the buyer is an ops lead at a company running hundreds of drones, and they do not have time for a tour
4
3
14
714
this is the quickest way to check if ai search engines can even find your product:
follow these steps exactly:
1. open an incognito browser, go to chatgpt or perplexity, ask it the exact question a customer would ask when they have your problem
2. check if your product actually gets cited as a source, not just mentioned in passing or not at all
3. if it doesnt show up, check yourdomain.com/robots.txt. see if gptbot, claudebot, or perplexitybot are blocked. most sites block these by default and nobody notices
4. confirm your key pages arent js-only rendered. ai crawlers dont execute javascript the way google's crawler does, if the content only loads after js runs, its invisible to them
5. pull your top 10 highest-traffic pages from analytics
6. paste one page's content into claude with this prompt:
"here's the content from this page: [PASTE]. does it answer one clear question in the first 2-3 sentences, in a way that could be lifted and cited directly? if not, give me the exact rewrite needed to make it extractable"
7. check those same pages for json-ld structured data. if theres none, add it, this is literally what tells a machine what the page is
8. link your blog content back to the actual product pages you want cited. orphaned articles rarely get cited even when they're good
5
2
13
733
Behind the scenes from another client build we embedded our automation for AEO/SEO
build out a seo automation workflow using ahrefs + dataforseo !
on track to hit 30,000+ impressions this quarter.
zero human in the loop after setup
- keyword filter: under 30 KD, over 1000+ monthly volume, nothing else gets in. anything above that gets auto-rejected before it reaches the queue ( for this phase )
- cluster the queries instead of writing one-off posts. one seed keyword spins out 8-12 related angles, all internally linked, all targeting the same intent ( scrapling integrations )
- research stage pulls real citations and live-verifies each one before it gets used. no invented stats, no dead links, no made-up sources. ( verified from dataforseo during qa )
- draft stage writes to a locked voice profile trained on actual past posts. same sentence rhythm, same structure rules, every single time.
- quality gate scores the draft against a rubric before it ever goes live. word count, structure, citation density, all checked in code, not guessed by the model.
- sounds harsh but the gate is the whole system. anything under threshold gets kicked back for a rewrite, not published anyway.
- publish stage runs on schedule, nobody touches it between draft and live. no editor, no approval queue, no bottleneck.
- 5 stages total, fully autonomous, same pipeline every time, no manual step anywhere in the chain.
2
2
7
1,009
This is not just happening in brand design.
founders are doing it to their products too !
This year we worked with a founder fresh off a $5M+ exit.
he wanted to start again, so we scoped an end to end ai recruitment platform with him and worked through the details , upfront scoping and kickoff
then kickoff came and we received literally a 15 page brief. written by GPT !
let me explain…
it read really well but man it had the vaguest pointers and so much random fluff
we told him it read like chatgpt. he did not deny it. he said he was trying to speed things up and keep everything articulate.
fair enough. so we went through all 15 pages, cut it down into a brief we could actually work with, and mapped out the full user journey ourselves with as much detail as we could.
then we sent it to him for review.
the review came back written by ai too.
here is the way i look at it though.
i do not have a problem with founders using ai. we use it all day to build. but the brief and the feedback are the two places where the founder's actual thinking has to show up. that is the part we can't generate for you.
so we paused and had the conversation.
I literally told him over the call
"every time you answer us through ai, even a small question, we get maybe 40% of what a handwritten answer from you would give us. and every one of those answers pulls the product around 10% away from what you had in your head"
do that across a whole build and we end up shipping something none of us pictured at the start.
so if you are about to hand your builder an ai written brief to save a week, you are saving time on the one part only you can do.
Got my first Ai slop Content Brief for a Brand design client. Please stop
I had to then give it to my agent to summarise it
3
1
5
764
backlinks we actually verify before submitting any dreamlaunch client site (v1):
1. findly.tools (dr 77) → dofollow, live same day
2. twelve tools (dr 82) → badge live
3. turbo0 (dr 79) → badge live
4. openhunts (dr 51) → badge live
5. easydofollow (dr 42) → badge live
6. similarlabs (dr varies) → badge live, review pending
7. indieascent → badge live, review pending
8. thesaasdir → badge live, review pending
9. toolfio → badge live, review pending
10. mochilaunch → badge live, review pending
11. sourceforge (dr 92-93) → no rel attribute at all on the outbound link
12. capterra (dr 91-93) → vendor profile
13. crunchbase (dr 89-91) → company profile field
14. wellfound / angellist (dr 89) → company profile
15. clutch.co (dr 90) → profile on signup
16. dev.to (dr 90) → profile website field, confirmed rel="noopener me"
17. startup fame (dr 82-83) → free "verified" tier, badge + dofollow
18. f6s (dr 83) → startup profile
19. stackshare (dr 80) → add to your dev-stack
20. toolpilot (dr 77) → badge
21. uneed.best (dr 74-75) → badge exchange
22. softwareworld (dr 73) → no approval queue
23. dofollow.tools (dr 72) → the badge is the entire premise
24. future tools (dr 69) → curated, ~75% rejection rate
25. opentools.ai (dr 68) → spot checked 68 live pages, 0 nofollow
26. betapage (dr 68) → upvote based
27. ai toolz dir (dr 63) → reciprocal backlink instead of badge
28. saasworthy (dr 52)
29. ventureradar (dr 51) → low traffic, clean dofollow
30. aitools.fyi (dr 44)
3
3
9
745
AGENT BUILDERS, this kills your worst bottleneck:
most agents dont fail on reasoning. they fail on data. you scrape, get blocked, patch the parser, repeat it next week.
TinyFish just launched a Data Partners Alliance with crunchbase, similarweb and 13 more companies.
licensed data, curated by category, pulled straight into your agent.
no proxy stack. no broken selectors.
tinyfish.ai/partners
AI agents are the web's biggest users.
They run on TinyFish, so we see what they reach for: data that's licensed, not published.
Today, with @crunchbase, @Similarweb, and 13 more companies, we're launching the TinyFish Data Partners Alliance.
The best data, curated by category, for agents.
tinyfish.ai/partners
Paid partnership (ad)
1
3
6
754
how we designed omegcol's onboarding system
> Inter Tight runs the UI, Lora serif carries headline copy. greeting and value slides sit at 26-29px, card titles 15px, button labels 15.5px, section labels 12-13px
> headline tracking sits at -0.3px so big serif type stays tight against the photo. section labels flip to +0.4px because uppercase micro labels read better opened up, not tightened
> no shared light/dark token set here, one dark photo backdrop drives every surface. white glass panels promote to a cream fill on selection, ink text stays constant throughout
> cards carry zero border color, a 1.5px ink ring marks the selected state instead, plus two shadow presets total, raised and raisedSoft, nothing tuned per component
> radius follows a hierarchy: 27-28px for the button pill, 20-26px for cards, 16px for chips and input rows, 12px for icon slots, 10px for the checkbox
> spacing stays dense, 20px page margins, 12px gaps between rows, 22-32px between step sections
once these tokens were locked the same button, same card, same type scale carried the whole onboarding flow, seven steps deep, nothing rebuilt per screen
steal this dark UI system we built for omegcol's onboarding
- headlines: Lora serif bold, 26px, -0.3px tracking, pure white over the photo
- intro slides go bigger, 29/37px, same -0.3 tracking, so the value props read like a magazine not an app
- card titles: Inter Tight semibold 15px, description 12.5/17
- buttons: 54px pill, radius 27, cream fill over ink text, never a flat dark button
- selected state is a border ring not a shadow, because white vs cream alone doesn't read as "selected" on a busy photo
- cards radius 20, chips radius 16, icon slots 36x36 at radius 12, nothing's a true circle except the checkbox
- progress track sits straight on the art at 4px tall, radius 2, a white card there would've killed the photo
- two shadow presets total, raised and raisedSoft, that's it, no per-component tuning
one photo backdrop, one type system doing two jobs, serif for voice and Inter Tight for UI !
1
2
5
793
Dreamlaunch completed almost 1.5 years with one founder based out of Norway
Took care of branding, web app, documentation and sometimes even helped them with pitchdecks for investor demo
It has started with a simple build of a event management platform idea
Then we kept scaling up feature releases alongside their marketing team. A lot of hustle went into infra, user experience and building it to be enterprise ready
slowly this entire collaboration became into feature additions and retainer's leading into :
- 750k+ site visitors
- 3.07m+ page views
- 15k+ registered users
- 4.7k+ paying users
Banger job done, now onto another year 🫡
2
8
707
steal this dark UI system we built for omegcol's onboarding
- headlines: Lora serif bold, 26px, -0.3px tracking, pure white over the photo
- intro slides go bigger, 29/37px, same -0.3 tracking, so the value props read like a magazine not an app
- card titles: Inter Tight semibold 15px, description 12.5/17
- buttons: 54px pill, radius 27, cream fill over ink text, never a flat dark button
- selected state is a border ring not a shadow, because white vs cream alone doesn't read as "selected" on a busy photo
- cards radius 20, chips radius 16, icon slots 36x36 at radius 12, nothing's a true circle except the checkbox
- progress track sits straight on the art at 4px tall, radius 2, a white card there would've killed the photo
- two shadow presets total, raised and raisedSoft, that's it, no per-component tuning
one photo backdrop, one type system doing two jobs, serif for voice and Inter Tight for UI !
3
2
5
1,339
Linkedin has been a big W platform for dreamlaunch.studio
I've been following a simple strategy
- personal branding
- behind the scenes
- authority posts
- awareness posts
In this past 365 days of posting, this brought in
- 7 projects kicking off
- 15+ prospect calls ( excluding unqualified )
- almost $100k in lead pipeline
- hitting almost 12,000 followers
- 3 major news features
Now onto scaling this more ! Confident I can hit 10M within 2027
3
3
10
1,047
Been watching @_anishkaran build this quietly for a while w/ hellosol.app
Most AI demos show i have seen are been usually about a chat window.
Sol shows the boring middle: the research, the doc, the slides, the calendar hunt that sits between an email request and the finished thing.
That middle is where my week goes.
Huge congrats on the launch. Excited to put a real task through it !
The obvious is missing.
So we built Sol - hellosol.app
Sol finds the work itself, does it, and comes back for your approval.
Every day in our emails we say "I’ll share”, "I'll review”, "I'll get back" - then repeat the exact same thing to an AI. Why?
Sol finds everything you said you’d do & gets them started for you.
It does the research, creates the doc, builds the slides, finds the time, connects the dots across multiple emails, doing everything it takes to get the job done - but doesn’t send, schedule, or share anything until you approve.
Sol runs on its own computer, uses a browser, and has a library of skills that automatically get assigned to the work that needs to get done. No setup. It just starts working.
We've raised $4M from General Catalyst, Nexus Venture Partners, DeVC, PeerCheque, Kunal Shah, and a few others.
Extending early access now.
@generalcatalyst @nexusvp @DeVC_Global @peercheque
@neerajarora @b_jishnu @kunalb11 @miten @RTinkslinger @Rahul_J_Mathur @AkarshS27 @SiddhantD06 @RajatAgarwal167
2
5
5
728
how we build products w/ agents that actually holds up at production scale :
everyone's obsessed with getting the demo working... thats the smallest part of the job
you need an actual system for scale, and thats the only way we take something from working demo to a product with real users on it :
> start from a checklist of what actually breaks at scale, slow queries, unindexed tables, no caching, no rate limits, this is your baseline before a single feature ships and it should get longer with every project, not stay static
> every time something breaks in production on a client build, that failure becomes a permanent rule for the next one, we don't just fix it and move on, we write down what caused it and why
> don't leave it as a "we should probably test this" note in a slack thread, build it into a pre launch checklist someone actually has to sign off on before we push to production
> feed the model the clients real expected load and data volume before it writes a single query, tell it how many users hit this in month one vs month six, not just the happy path with 10 fake rows
> stress test with real numbers before real users ever touch it, simulate the traffic spike, not the average day, most bugs only show up when the system is under pressure
> build monitoring and alerting in from day one, so you catch whats breaking at 2am before the client wakes up to angry emails from their own users
> separate what the ai wrote fast from what actually needs a human to slow down on, auth, payments, and anything touching user data always gets a manual pass no matter how confident the model sounds
> version everything the ai touches so a bad generation is a five minute revert, not a three hour debugging session trying to figure out what changed and why
Been running the same workflows for the past year
3
4
7
672
Here's a 30-second Product tip :
add an "interactive demo" to your product home page to win prosumers
> let people click through it:
- do = a real, or realistic, version of the product
- don't = a 90 second video explaining features nobody asked about
> show the real thing:
- do = the actual dashboard/data view if your product has one
- don't = a static screenshot with fake blurred numbers
> keep it narrow:
- do = one core flow, the thing that sells the product
- don't = try to demo every feature at once
> let people leave and come back:
- do = let visitors leave the demo and return to it
- don't = gate it behind a signup wall immediately
do these 4 things, and your home page sells the product before anyone talks to sales.
don't overcomplicate it !
1
3
7
741
Bro this is CRAZY !
Deel just added 140M added to AR using Akay
Most people write a prompt to automate their work.
Deel presents a better way: record yourself doing the job once and let the agent write the spec.
What that actually captures from a real billion dollar workflow:
→ the steps you'd never think to write down, like checking legibility before amount
→ the ordering rules you follow without realizing it, like country first, then fields
→ the exceptions you handle on instinct, like splitting a bulk receipt into separate lines
→ the systems behind each click, so the agent can work through the same tools you do
Deel used this approach to take expense review from 40+ hours a month to under 1 hour
The team processes 10,000+ expenses a month, and the workflow was built in under 2 hours by someone who doesn't code
That's a much more interesting way to automate work than writing another prompt.
EXCITED TO LAUNCH: Akai (akai.run)
Deel added >$140M ARR in 90 days without increasing headcount by automating~600 Full Time Employees' equivalent in work with Akai.
Akai was an internal tool to automate our painfully repetitive operations in Finance, HR, Accounts Payable, and Compliance, etc. We never intended to make this a product.
But we watched revenue per employee grow from $130K to $215K
We built >8k agents that do the work of ~600 employees
It had such a dramatic impact on our business that today we are launching it for everyone.
How it works: Say you're automating payment reconciliation:
1. Record your screen while manually matching a messy transaction and Akai will capture your screen, voice, server requests
2. Akai will see that you pulled unformatted wire transfer info from an archaic bank portal, put it in some excel sheet, checked NetSuite invoices, payment history, and put a ticket on Zendesk
3. Akai reads between the lines and build a workflow + steps + conditional guardrails. It learns tacit edge cases, like resolving malformed invoice references without you writing a single regex
4. Simply connect NetSuite, your ledger, Zendesk, PSPs, and even legacy bank portals with zero API access
5. Run the workflow and tell it what to adjust in plain English: "strip slashes on wire memos and auto-apply partial payments." It adapts instantly
6. Once it works for you, add 100s of colleagues. Your entire payment ops team forks and extends the workflow for new PSPs, secondary ledgers, or regional settlement rules
7. We automated 85% of our payment reconciliation end to end, eliminating 500+ hours of soul-crushing manual grunt work every single week.
Claude Code/Codex can't do this in multiplayer mode. Every person rebuilds the same skill from scratch in their own way.
Deel built Akai to:
1. Understand backend operations edge cases (it had to work for our 7000 person team first)
2. Collaborative across 1000s of employees
3. Self-Learning from millions of runs
4. Optimises cost and gets cheaper every run
We're so confident that we're announcing an Automation Guarantee:
If our engineers can't automate a thousand of hours of work in your first 30 days, you get a full refund.
Book a demo: akai.run if you're an exec at a company with hundreds of employees
Paid partnership (ad)
2
7
759
add "real case studies" to your product to win enterprise buyers
here's a page out of the case study we built for tenmo, one of our clients:
1. lead with the outcome in the headline, not the client's name
do = "cutting incident response time 38% across a 12,400-drone fleet"
don't = "how [client] uses our product"
2. pair every stat with its before number
do = "-38%, down from a 22-minute average across five separate dashboards"
don't = a bare percentage floating with nothing to compare against
3. anchor it with specifics a buyer can benchmark against their own org
do = industry, hq, plan tier, exact modules deployed, listed as a strip under the headline
don't = "enterprise client" with no scale or scope attached
4. show the rollout, not just the result
do = week 1 connected, month 1 onboarded, month 6 scaled 4x with uptime holding steady
don't = one results number with no proof it survived contact with real usage
5. close with a quote from the exact role making the buying decision
do = "fleet operations, amazon prime air"
don't = an anonymous "happy customer" quote
do these 5, and the case study becomes something a buyer forwards internally instead of skims and forgets.
don't overcomplicate it, one real number beats ten adjectives.
2
3
10
719
Funnily enough I had planned to be a digital artist, but ended being into computer science
i remember i saved up almost $100 to buy my first drawing tablet and asked my friend to amazon gift me ( told my parents i won it in a giveaway )
then i kept working on the tablet and did some freelance gigs to finally purchase a $400 huion drawing tablet
but i got burnt out and wasn't able to see the results. so i switched to design ( UI UX ) which eventually got me curious about frontend development
that whole transition lead me to this roller coaster journey of going from on-campus placement to a remote job to building dreamlaunch.studio
if i could go back and tell the little harshil, you would be doing AGENTS, and building global products for Clients, he would be so confused
truly : life is one BIG GAME, life is CRAYZ !
i was supposed to go into finance, but ended up being in computer science lol
during covid i used to play minecraft with some online friends on discord, i used to see them make crazy stuff back then
got curious and started making mods, game plugins and discord bots with them
all this was pre AI, pre everything
someone then paid me $35 to fix a button in html, change some text and fix css
it was then, when i realized.... like whoaaa
i never even thought i'd take this as a career, like NEVER, it was just super interesting
all this because of a GAME, life is CRAZY
4
3
10
1,009