Case Study
How AcadAlly Added Over 6 Figures to Net Profit with Claude in 3 Months
Inside a 4-team AI rollout across customer success, content, engineering, and product and design, with Serviceful as AcadAlly’s integrated AI team.
Case Study
Text LinkIntroduction to AcadAlly
AcadAlly is a leading K12 Learning Management System (LMS) and EdTech SaaS platform, helping K12 schools run AI-driven adaptive learning and assessment. The company wanted to adopt AI for revenue acceleration and employee productivity. A rapidly growing startup present across 28 cities and adopted by schools like DPS, GD Goenka and Lotus Valley, AcadAlly needed to deploy technology in the right places of the business to support its ambitious plans. While a few employees were already using AI tools such as Claude and ChatGPT for ad-hoc work, adoption was inconsistent across the team. The company needed an overarching strategy that tied business goals back to processes, tooling and AI.
With over 50 employees and growing at 100% YoY, AcadAlly had a number of IT systems already in place, including a CRM, ticketing and source control, but needed to scale its operations while making bold bets on future-proof technology that could provide meaningful long-term financial gains.
Results at a glance
A leading EdTech platform serving K12 schools

has successfully completed its cross-functional AI transformation across 4 teams
Powered by

Measured a net profit improvement of
The Process
Serviceful was brought in to play the role of an integrated AI team to deliver on 3 things: 1) audit and recommend opportunities, 2) deliver Proofs of Concept, and 3) implement solutions alongside existing team members and track performance impact.
Using our ROIC methodology, we identified more than 18 use cases for AI adoption across business functions, covering customer success, product and engineering, educational content and more. We then prioritized the ones with the fastest payback across 4 teams:
Use cases delivered
18+ opportunities, narrowed to the 4 teams where AI would pay back fastest18+ opportunities, 4 priority teams
Customer Success
Tier-1 support automation
Claude answers routine school questions automatically and routes the rest to account managers.



Educational Content
Question generation and translation
Claude drafts questions and translations, and SMEs review and sign off.

Engineering
AI-assisted development and QA
AI coding tools for every engineer, plus automated regression testing.


Product & Design
PRDs and design prototyping
Faster specs, research and prototypes from one shared toolset.


Each use case is mapped to a node on AcadAlly’s ROIC (return on invested capital) tree.
Challenge
One of the biggest gaps identified upfront by their management was tooling fragmentation and confusion. There was a lack of clarity on which tool to use for what purpose. For instance, a few members of the engineering team had bought Cursor and Claude Code licenses on the company card, while others on engineering were sticking to GitHub Copilot in VS Code. Similarly, a member of the customer success team was actively using Claude Cowork to create reports for school administrators, but was doing it in isolation without the rest of the team benefitting from that process.
At the leadership level they had a CTO, but didn’t have anyone to own AI systems across business functions to review governance, budgeting and adoption. They needed someone to define the new process and workflow using the latest technology.
As a team of Claude Certified Architects with a diverse systems understanding, Serviceful was made responsible for defining the right process and AI tools that their current teams could make use of today.
Solution
We ran the engagement from May to July 2026, working on-site with AcadAlly’s teams. Rather than starting with tools, we started with the business: which processes mattered most, what they cost today, and where AI could move a number the team already tracked. That is why the priorities landed in Customer Success (CSS), Educational Content, Engineering and Product & Design: the 4 teams where AI could move a daily metric within weeks.
This is where our ROIC methodology came into play. Every use case was tied back to a node on AcadAlly’s return on invested capital (ROIC) tree, such as Operating Costs or R&D & Overhead, along with its expected impact on the bottom line. That gave the founders one view of which bets would pay back fastest, so they could fund those first instead of buying tools team by team. Each team then signed up to 1 or 2 KPIs it already tracked: questions created per day for Content, tasks completed per sprint for Engineering, turnaround time and manually handled tickets for CSS, and PRDs and designs created for Product & Design. Baselines were captured before anything was rolled out.
The rollout itself started deliberately small. Phase 1 put 10 Claude licenses, covering Claude Code and Claude Cowork, in the hands of the people closest to the priority use cases, with a shared API budget and Cursor and OpenAI licenses only where a team had a clear need. After 2 weeks, we reviewed utilization and moved licenses to the people getting the most value from them. Alongside this, we ran in-person, hands-on trainings over 6 weeks using each team’s own tickets, question banks and specs rather than generic demos, so every session ended with something the team could use right away. Every team member also has ongoing access to our training materials on Claude Code and Claude Cowork, so new joiners can get up to speed without waiting for the next session.
Most importantly, we didn’t replace systems and processes that already worked. We started from how each team operated and added Claude where it removed the most manual effort.
In Customer Success, every partner school has a WhatsApp group where teachers ask questions. We set up a scheduled Claude task that reads every thread every 30 minutes. Claude answers 80% of questions automatically, right in the thread, while anything more involved becomes a ticket in Odoo for the Key Account Manager (KAM) to resolve. Account managers still read along in every group, so there is always human oversight.
Customer success workflow
Claude answers 80% of school support questions automatically

Teachers ask
in their school’s group

Reads every thread
every 30 minutes
80%
Claude answers automatically in the thread
KAM resolves the ticket
Human oversight
Account managers read along in every group.
In Educational Content, every LMS question now starts from a standard Google Doc learning unit (LU) template used internally. The template captures curriculum mapping (topic, learning unit and learning outcome), Bloom’s level, difficulty, the question stem, 4 options with the distractor type and rationale for each, a hint and a worked solution. Claude drafts against the exact template and an SME reviews and signs off, which keeps quality consistent and makes review much faster.
Assessment content workflow
One question template behind 50% more SME-approved questions every day
For Engineering, Product and Design, we replaced the patchwork of licenses with one clear stack per team. Engineering uses Claude Code and Cursor, backed by an API budget for Claude Code, and uses Claude to automate QA regression testing. Product uses Claude Cowork for PRDs and research, while Design pairs Figma AI with Claude Cowork for prototyping, with individual Claude licenses for every product and design team member.
R&D AI tool stack
The right AI tool for every team, leading to 70%+ more tasks per sprint



AI
Engineering
Generous API budget for Claude Code
Product & Design
Individual Claude licenses
70%+
More tasks completed per sprint
Measured over 3 months across engineering, product and design
The audit also surfaced several new AI-led product lines, which are now under development. The details are confidential for now, but more on that soon.
Outcomes
Within 3 months, all 4 priority teams moved on the KPIs they had signed up to, and AcadAlly measured a net profit improvement of over 6 figures (USD), driven by lower operating costs and more output from the same team.
In Customer Success, Claude now answers 80% of school support questions automatically in WhatsApp, leaving KAMs free to focus on the issues that need them. As a result, each KAM now manages 30 schools instead of 20, a 50% increase in capacity. The setup went live within days, with barely any help from engineering.
“Setting up Claude with our support tools felt like it would take weeks. It came together in a few days, with barely any help from engineering. Within 2 weeks, our account managers were responding faster and spending less time on repetitive replies.” Ankit Tripathi, Customer Success Manager
The content team, AcadAlly’s biggest and supported by outsourced freelancers, saw the most direct productivity gain. SMEs moved from writing every question from scratch to reviewing and editing Claude’s drafts, and SME signed-off questions per day for Science & Math went from 30 to 45, a 50% increase in output across Grades 5 to 10. Language translation, traditionally slow and expensive, is now quick and costs a fraction of what it used to, opening up new geographies without adding SMEs.
Across Engineering, Product and Design, PRDs are now prepared 3x faster, and tasks completed per sprint are up more than 70%, measured over 3 months across development, design prototyping and code generation.
Beyond the 4 teams, the biggest change was at the top, in the quality of decision making. The founders now use Claude as a thought partner on hiring, competitive analysis and fundraising, and to turn strategy into board-ready material in hours instead of weeks.

AI transformation outcomes
Process first, then Claude: 4 teams measurably better in 3 months4 teams better in 3 months


Foundation
Processes defined across existing tools
Customer success
KAM to school ratio
1:20 → 1:30
+50% capacityEducational content
Questions per day, SME signed off
30 → 45
+50% outputEngineering, Product & Design
Tasks completed per sprint
+70%
Bottom line
Net profit improvement
Over 6 figures
USD, within 3 monthsBonus: new revenue
AI product lines identified
Now under development
Bonus: leadership
Sharper strategic calls
Hiring, competitive analysis, fundraising
Process first, then Claude AI
The lesson from AcadAlly is simple: process first, then Claude AI. With processes defined, one clear tool stack per team and a KPI each team owns, AI stopped being something a few people used on the side and became part of how the company runs, and the gains showed up where they matter most: the bottom line. That foundation is what AcadAlly’s next AI product lines are now being built on, unlocking new revenue opportunities for the company.
Your business next
Find where AI pays back fastest in your business
In a 30-minute call, we’ll map your highest-cost processes to where Claude can move a number you already track.
Book an AI opportunity callRelated articles




