Why Agents (and not Chatbots) Will Deliver Software's Value Promise
Software has an epidemic problem that few have spoken about and dare to acknowledge: despite investing in sophisticated software solutions, teams consistently gravitate back to Excel ( if you swing for gsuite then Google Sheets). I have seen this for orgs of 5 people, 50 people and 500 people. This isn't just an anecdote – it's a systemic issue that reveals deeper truths about software adoption and value realization.
The Excel Paradox
If you have worked in SaaS, or just used software, you have probably experienced this too: Sales teams abandoning CRMs to manage their reviews in Google Sheets, Engineering teams side-stepping Jira for sprint discussions, returning to – you guessed it – spreadheets, Marketing teams forsaking dashboards for the familiar comfort of ... Google Sheets, Finance teams chasing employees to use the latest expense management software only to get emails with Google sheet links. This list is endless.
For the longest time, I looked at this as an enablement problem only to conclude that it's not. The spreadsheet's persistence isn't a bug – it's a feature of human behavior and organizational dynamics. You'll always find power users within an organization. The real challenge? Getting everyone around you to use the same software, consistently - because that's when value gets realised. It's not just hard; it's super hard. Getting people to actually use software is exponentially harder and no sales rep will tell you this when selling.
The SaaS Paradox, Conversational AI Promise And A Reality Check
When GPT-3 emerged two years ago, many of us hoped GenAI would finally solve this adoption crisis. We've heard various narratives: Outcomes-based pricing / Service as Software / Usage-based pricing / The death of traditional SaaS metrics
Yet until earlier this year, what did we really get? GenAI Chatbots. Lots of them. Layered with versions of RAG. There are now 20+ versions of RAG out there. Don't get me wrong - chatbots are a significant upgrade and they solve a ton of problems, atleast on the CX front. GenAI chatbots are loaded with impressive conversational abilities, but for most part, they are merely a new interface layer loosely integrated with systems of record. Sprinkling AI with a chatbot in your product, hoping that it would fix your adoption problem is just passing the monkey to someone else. The chat ux is not the silver bullet it was sold as.
Only some people actually use chatbots. Most just play with it and go back to doing what they were doing.
The results speak for themselves: CIOs are sitting on Microsoft Copilot licenses with usage hovering in low single digits.
Software for the longest time has been built for a type of organizations, or type of users. We have sold, and bought one-size fits-all software and that only gets us so far. The Agentic approach promises us - software for an audience of one.
AI Agents to Build Software For An Audience of One- A New Silver bullet or A Better Lead Bullet?
Agents promise customizing software for an audience of one. It's the hyper-personalization that promises to solve the problem of adoptiom. Agentic systems promise to:
领英推荐
Customer support, a popular use case is served both by the Chat UX and by agents that draft responses, process refunds/ returns / update CRM and status and close the loop.
Curosr IDE is another excellent example here. Recommends changes, applies them to code, helps you debug and deploy - all of that using a chat based UX again. The end impact being 10X development pace.
Agents are not a silver bullet but you definitely won't need a 100 of those lead bullets now for the same problem. Jargons aside - Agentic systems are quickly evolving from Software for a category to software for actual jobs to be done.
This is also why the tide is turning towards agents and agentic systems.
Passing The Adoption Monkey To Agents
Given the pace of improvement and lift in productivity, we are definitenly looking at change of how "work gets done". For decades, we've focused on making software more user-friendly, adding features, and creating better interfaces – all in service of getting humans to adopt and use these tools. But perhaps we've been solving the wrong problem.
If we are actually able to pass the adoption baton to agents, here are some questions to think about -
May be the key to breaking the spreadsheet cycle isn't building better interfaces – it's possibly eliminating them altogether by creating systems that actually do the work for us.
How are you thinking about agents and agentic systems?
Entrepreneur | IT Consultant | TedX Speaker
2 个月You very rightly pointed out the excel paradox. While I always presumed it was ease of convenience or merely habit that led people to obsessively use excel. I enjoyed reading your take on AI Agents for this scenario.
Building & Deploying Agents For Enterprises | Applied AI Services
2 个月In a 90 min conversation, Satya Nadella shares about how he uses applications and confesses how most application UI will collapse -https://open.spotify.com/episode/640nHdVvWgOp8tYShOpXH1
Co-founder @Sherlocks.ai | Turning AI into your on-call sidekick ? ex-CTO at Doubtnut - India’s Largest Education App | Co-founder at Aasaanjobs (acquired by OLX)
3 个月Wow! The Spreadsheet line of thought hit hard! Never thought of it this way. Well articulated Vivek Khandelwal!
Building & Deploying Agents For Enterprises | Applied AI Services
3 个月Here's a first glimpse of what Agents will do for us - https://x.com/joshm/status/1863580629465788823
Nuvepro = Project Readiness
3 个月In future, every software maker will have to have two interfaces - a human one, and an agentic one. A Human one to close a commercial deal basis the capabilities of the software, and an agentic one which will get utilized to perform tasks. Every company will run with a Agent OS that will interact with the agents of software that this company has purchased.