Plain English Is The New API

Reston, Virginia. On 7 July, an American technology firm called Peraton launched what it billed as the first true enterprise agentic AI platform for government operations. In the report in NextGen Defense, the description of the tool is deceptively simple:

Users can query the system in plain English to identify project risks, monitor progress, and gain real-time insights.

I read that sentence twice. Not for the marketing gloss, but for the quiet implication buried in it.

For three decades the story of government IT has run the same script. Buy an enterprise system, spend two years customising it, train a small priesthood of operators, live with the quirks for a decade because migration is unaffordable. The bottleneck was never data. It was the specialist layer between the user and the data. Any officer who has ever needed a report from a legacy application and been told we will raise a ticket knows this bottleneck in her bones.

If the plain-English promise even half holds, that layer starts to thin. A field officer who wants to see all pending appeals in one district by tax head, or the desk officer tracking anomalous refund patterns this quarter, would ask the system directly. No ticket, no intermediary, no six-week wait.

The catch, and it is a serious one, is traceability. In administration, the model said so is not a defensible answer. Every output that touches a decision must tie back to a rule, a return, a scrutiny note. Vendors are already promising this loudly. Governments will have to test it just as loudly, on their own data, in their own languages, with adversarial cases picked by their own auditors.

A modest proposal for any large Indian department contemplating agentic AI. Insist on three non-negotiables inside the procurement itself. First, an offline sandbox on real, redacted departmental data before any commitment is signed. Second, a written explanation for every query result, citing the source records. Third, a full audit log that a Comptroller can read a year later without help from the vendor.

The novelty here is not the model. It is the interface. When plain English becomes the query language, the constituency for institutional data widens from the few hundred people who know the schema to every officer with a question. That is either a productivity revolution or a governance nightmare, depending entirely on how quietly the audit trail is built.

#AgenticAI #PublicSectorAI #GovTech #IndiaGovernance #DigitalGovernment #TaxAdministration #AIProcurement

Stop Building Chatbots.

Eighty-two percent. That is the share of inbound citizen queries an AI agent called Bobbi resolved in its first week across three English police forces, without a single hand-off to a human officer. The number matters less for policing than for what it telegraphs to the rest of government: the chatbot era, finally, is ending.

I say this with some scepticism about my own past. Anyone who has helped stand up a citizen-facing assistant inside a large national department knows the temptation to call any conversational widget a chatbot and declare modernisation complete. It is not. Bobbi, alongside Singapore's GovTech work and Estonia's interoperable agent network, signals that the next layer of public-sector AI will look very different from what most administrations are currently buying.

Chatbots answer. Agents act.

The technical distinction is sharper than the marketing suggests. A chatbot replies to a prompt. An agentic system is handed a goal — process this permit renewal — plans the steps, calls the databases, validates against the rules and executes the transaction. One produces text. The other completes a workflow.

This is why a recent World Economic Forum and Capgemini exercise mapping seventy core government functions sorts them more cleanly by workflow than by department. Eligibility assessment, document processing, fraud detection, permit issuance — these cut across ministries; they are the natural unit of analysis for an agent, not an org chart. Agents do not need to break silos. They operate outside them.

The chatbot habit is the trap.

The single most common mistake I see, both in India and abroad, is treating an agentic deployment as a chatbot upgrade. Same procurement template, same vendor, same knowledge base, a slightly cleverer model bolted on the back. That is not what the technology is for. The Capgemini survey of 350 public-sector organisations finds that ninety percent intend to explore or deploy agentic AI within two to three years, while Gartner forecasts that more than forty percent of those projects could be cancelled by 2027 — usually because the agency moved before understanding where the actual value sits.

The value sits in outcomes, not interactions. The right question is not can the bot answer this but what is the citizen actually trying to finish, and how many steps can we collapse into one supervised flow. An estimated one hundred and forty billion dollars in US federal benefits goes unclaimed each year because the application paths are too fragmented for the people who most need them. Most large administrations, India included, will find similar pockets if they look honestly: schemes whose stated reach far exceeds their actual delivery, not for want of policy but for want of plumbing.

What conversational AI for citizens actually teaches.

Citizen-facing conversational systems at national scale teach two unfashionable lessons. First, the volume of routine queries is far larger than any budget anticipated, and far more repetitive: a small set of questions accounts for most of the load. Second, when the citizen needs to complete something rather than learn something, the conversational layer hits a wall. The handover to forms, portals and back-office staff is where the experience breaks.

Agents are precisely the technology for that gap. In a direct-taxes context, an agent can read a notice, retrieve the relevant return, pre-fill a form against rule sets, cross-check historical filings and route only the genuinely ambiguous cases to a human officer. The administrator stays the architect. The agent does the clerical labour that nobody, on either side of the counter, particularly enjoys.

Bounded autonomy, glass-box defaults.

The discipline the field is converging on is bounded autonomy: being deliberate about what an agent is allowed to do, keeping a human meaningfully in the loop, and making every step auditable. The phrase that fits a public-sector context is glass-box governance. A chatbot answers, and the trail is shallow. An agent acts, and each action — every rule consulted, every database queried, every form submitted — should leave a perfect record. Used well, this is more accountable than the human-only system it replaces, not less.

The procurement template should change accordingly. Stop buying chatbots. Buy a workflow agent that ships with an audit log by default, with explicit, narrow permissions per task and a hard escalation rule for any case touching rights or material amounts. Bounded autonomy belongs in the contract, not in a vendor promise.

Where to start.

The most useful first move for any Indian department is unglamorous: pick one workflow that already exists end-to-end on paper or in disconnected portals — refund issuance, grievance redressal, a single notice-and-reply cycle — and rebuild it as an agentic flow with a human checkpoint at the decision. Measure completions, not interactions. Compare cost not to the chatbot it replaces but to the staff hours it returns. That number is what budget committees will eventually demand, and it is the only one that matters.

The chatbot was a useful detour. It taught millions of citizens that they could speak to the state in plain language and get a sensible answer. The agent is what turns that conversation into a completed transaction. The departments that grasp the difference now, and procure for it accordingly, will spend the next decade doing more with steadier headcount. The rest will spend it explaining why their bot still cannot actually do the thing.

#AgenticAI #PublicSectorAI #DigitalGovernance #GovTech #CitizenServices #AIinGovernance #IndiaAI

Agentic AI in Banking

Artificial Intelligence is evolving rapidly, and its newest frontier, known as Agentic AI, is beginning to reshape how the banking industry operates. Unlike earlier AI models that merely responded to prompts, agentic systems are designed to act with autonomy. They can initiate actions, complete multi-step processes, and make decisions without constant human oversight. This shift promises enormous efficiency gains for banks but also raises significant concerns around security, ethics, and governance.

The appeal of agentic AI lies in its ability to transform routine operations. Already, around 41% of organisations in Australia report using some form of agentic AI and that's a huge number, and by 2029 it is projected that nearly 80% of customer service issues could be handled autonomously (Banker's jobs??). For banks, this translates into faster response times, reduced costs, and an enhanced customer experience. Compliance monitoring is also evolving, with solutions such as Proofpoint’s Human Communications Intelligence claiming to interpret human conversations including slang, shorthand, emojis, and tone in real time. By moving away from keyword-based monitoring and towards contextual understanding, such tools claim up to a 90% reduction in false positives, which is a significant leap in operational accuracy.

Yet the same technologies that empower banks can also empower adversaries. Cybercriminals are increasingly turning to AI to sharpen their attacks, with as many as 80% of ransomware campaigns now incorporating AI-driven elements such as advanced phishing, social engineering, and deepfakes. The banking sector finds itself in the middle of an arms race: as institutions use AI to safeguard systems, attackers use the same technology to evade detection. At the same time, the rise of real-time monitoring of employee and customer communications introduces new ethical and regulatory dilemmas. Constant surveillance can conflict with privacy laws, employee rights, and expectations of confidentiality. Moreover, AI systems often struggle with cultural nuance, sarcasm, or intent, raising the possibility of misclassifications that could result in compliance errors or reputational damage.

There are also risks inherent to the AI systems themselves. If compromised, these tools could be manipulated to miss threats, conceal fraudulent activity, or even leak sensitive communications. This makes the integrity and security of the AI a crucial issue in its own right. As banks adopt these technologies, they must not only consider how AI can be used to detect misconduct but also how to protect the AI from becoming a new point of vulnerability.

The adoption of agentic AI therefore cannot be viewed purely as a technological upgrade. It must be accompanied by robust oversight and governance. Human-in-the-loop systems will remain necessary for sensitive or ambiguous cases, and transparency will be vital so that regulators and customers alike can understand how AI-driven conclusions are reached. Regulatory frameworks will need to evolve to demand clearer auditing, accountability, and disclosure around AI use. For global institutions, harmonising standards across jurisdictions will be essential to avoid compliance conflicts.

Agentic AI is poised to redefine the future of banking security. If managed wisely, it offers the potential to enhance efficiency, protect assets, and build stronger customer trust. But if deployed recklessly, it could amplify the very risks it seeks to mitigate, fueling fraud, misinterpretation, and data misuse. 

The challenge for banks, regulators, and technologists is not whether to adopt agentic AI, but how to govern it responsibly. Efficiency and security must grow together, or the sector may find itself solving one problem only to create another.